POPULARITY
Categories
308 episodes with Batch
170 episodes with Batch
44 episodes with Batch
70 episodes with Batch
56 episodes with Batch
59 episodes with Batch
57 episodes with Batch
25 episodes with Batch
39 episodes with Batch
26 episodes with Batch
20 episodes with Batch
70 episodes with Batch
22 episodes with Batch
19 episodes with Batch
19 episodes with Batch
82 episodes with Batch
41 episodes with Batch
22 episodes with Batch
37 episodes with Batch
11 episodes with Batch
12 episodes with Batch
15 episodes with Batch
24 episodes with Batch
16 episodes with Batch
9 episodes with Batch
18 episodes with Batch
12 episodes with Batch
22 episodes with Batch
15 episodes with Batch
12 episodes with Batch
14 episodes with Batch
19 episodes with Batch
19 episodes with Batch
13 episodes with Batch
13 episodes with Batch
14 episodes with Batch
12 episodes with Batch
17 episodes with Batch
8 episodes with Batch
![Engenharia de Dados [Cast]](https://ivyfm.s3.amazonaws.com/i320/1212346.jpg)
10 episodes with Batch
21 episodes with Batch
14 episodes with Batch
25 episodes with Batch
19 episodes with Batch
10 episodes with Batch
18 episodes with Batch
7 episodes with Batch
7 episodes with Batch
10 episodes with Batch
12 episodes with Batch
8 episodes with Batch
5 episodes with Batch
7 episodes with Batch
5 episodes with Batch
7 episodes with Batch
7 episodes with Batch
8 episodes with Batch
8 episodes with Batch
9 episodes with Batch
6 episodes with Batch
14 episodes with Batch
14 episodes with Batch
14 episodes with Batch
7 episodes with Batch
7 episodes with Batch
5 episodes with Batch
12 episodes with Batch
8 episodes with Batch
5 episodes with Batch
7 episodes with Batch
4 episodes with Batch
5 episodes with Batch
16 episodes with Batch
6 episodes with Batch
5 episodes with Batch
5 episodes with Batch
7 episodes with Batch
12 episodes with Batch
5 episodes with Batch
9 episodes with Batch
7 episodes with Batch
7 episodes with Batch
4 episodes with Batch
7 episodes with Batch
6 episodes with Batch
6 episodes with Batch
13 episodes with Batch
14 episodes with Batch
8 episodes with Batch
6 episodes with Batch
5 episodes with Batch
6 episodes with Batch
7 episodes with Batch
4 episodes with Batch
7 episodes with Batch
6 episodes with Batch
4 episodes with Batch
14 episodes with Batch
5 episodes with Batch
5 episodes with Batch
7 episodes with Batch
5 episodes with Batch
5 episodes with Batch
10 episodes with Batch
6 episodes with Batch
12 episodes with Batch
5 episodes with Batch
6 episodes with Batch
4 episodes with Batch
5 episodes with Batch
5 episodes with Batch
5 episodes with Batch
4 episodes with Batch
3 episodes with Batch
10 episodes with Batch
5 episodes with Batch
4 episodes with Batch
12 episodes with Batch
10 episodes with Batch
5 episodes with Batch
4 episodes with Batch
5 episodes with Batch
5 episodes with Batch
2 episodes with Batch
4 episodes with Batch
3 episodes with Batch
3 episodes with Batch
5 episodes with Batch
8 episodes with Batch
3 episodes with Batch
12 episodes with Batch
5 episodes with Batch
4 episodes with Batch
4 episodes with Batch
6 episodes with Batch
4 episodes with Batch
4 episodes with Batch
3 episodes with Batch
4 episodes with Batch
3 episodes with Batch
5 episodes with Batch
5 episodes with Batch
6 episodes with Batch
6 episodes with Batch
3 episodes with Batch
5 episodes with Batch
4 episodes with Batch
5 episodes with Batch
4 episodes with Batch
5 episodes with Batch
2 episodes with Batch
3 episodes with Batch
4 episodes with Batch
9 episodes with Batch
5 episodes with Batch
5 episodes with Batch
3 episodes with Batch
2 episodes with Batch
11 episodes with Batch
6 episodes with Batch
2 episodes with Batch
4 episodes with Batch
4 episodes with Batch
70 episodes with Batch
9 episodes with Batch
2 episodes with Batch
4 episodes with Batch
3 episodes with Batch
3 episodes with Batch
4 episodes with Batch
3 episodes with Batch
Hour 2 of JJ & Alex with Jeremiah Jensen and Alex Kirry. Eric Spyropoulos, digital writer for NBA.com and Utah Jazz NFL Blitz: Latest batch of NFL Report Cards gets leaked The Top 10: Least Valuable Sports Franchises
When Brown-Forman announced they would be releasing a new small batch version of King of Kentucky, we were excited and also had many questions. Instead of more single barrels, this remixed version of the "King" comes in the form of three batches with a brand new mashbill. For the first time, King of Kentucky has released a three-batch series, each bottled at a distinct proof to showcase the complexity of its aging process: Batch 1: 105 proof; Batch 2: 107.5 proof; Batch 3: 110 proof. For this episode, our friend Jason was able to secure Batch 2 and we supplied some blinds to make sure that the very first time we tasted this whiskey would be blind. How did things end up? Is this new version of the king really the "King?" You'll have to listen to find out. --------------------------SocialsIG: https://www.instagram.com/themashupkyFB: https://www.facebook.com/themashupkyYouTube: https://www.youtube.com/@themashupkyJoin our community on Patreon: https://www.patreon.com/TheMashUpBourbonPodcastPartnership(s)Visit Bourbonoutfitter.com and enter code THEMASHUP for a special discount or visit bourbonoutfitter.com/THEMASHUPMusic: All the Fixings by Zachariah HickmanThank you so much for listening!
What if the reason you’re not healing isn’t that you need another diagnosis? 0:08 It’s that your cells aren’t receiving the right signals. Because the body doesn’t run on diagnosis, it runs on 0:16 communication. And peptides are one of the most powerful, most misunderstood 0:21 tools we have for cellular signaling, immune balance, tissue repair, gut 0:27 lining support, metabolic control, brain signaling, sleep cycles, and even sexual 0:35 wellness. Today, I’m going to do what most people won’t. Define peptides in 0:41 plain English for you. break them into categories by what they’re best at and 0:47 tell you which ones are FDA approved on the list and which ones are commonly 0:53 used off label or investigational with the evidence that actually says these 1:00 work. This is going to be a powerful episode and if you’ve ever felt like you’re hearing hype without clarity, 1:07 this one’s for you. So, as usual, grab your cup of coffee or tea and settle in 1:13 as we talk about peptides that can fit into your healing journey. We’re going 1:19 to have a short word from our sponsor. You know, we got to do that. That’s how we stay on the air here. So, we will be 1:26 right back after this. Did you know sweating can literally heal your cells? 1:32I nfrared saunas don’t just relax you. They detox your body, balance hormones, 1:37 and boost mitochondrial energy. I’m obsessed with my health tech sauna. And 1:42 right now, you can save $500 with my code at healthtechalth.com/drmuthqen25. 1:54 All right, here we go, guys. I am excited to dive into peptides with you. 2:00 So understanding peptides is foundational, right? And I’ve been 2:06 studying peptides now for about nine years. Um, and I find that they are 2:13 incredible. Um, so I want to break down for you what peptides actually are, what 2:19 they do, and some of the top peptides that are available today, and how they 2:25 can be utilized. Because I think it’s really important. And I think it’s it’s there’s a lot of confusion out there about what these things actually are and 2:32 are they safe? Are they not? When do we use them? What’s the science behind them? So, we’re going to dive in and 2:38 we’re going to talk about all things peptides. So, let’s get ready here. Here we go. So, peptides are short chains of 2:45 amino acids and they typically range anywhere from 2 to 50 amino acids and 2:51 they’re linked by peptide bonds. So think of them as the superglue that holds the amino acids together. They sit 2:58 between the amino acids and they are full proteins in terms of their size and 3:04 their complex structure. And what makes peptides particularly interesting in 3:10 medicine is their role as signaling molecules. They’re essentially the 3:15 body’s text messages carrying specific instructions to cells and tissues. And 3:21 unlike our proteins which often serve as structural roles or act as enzymes, 3:28 peptides typically function as hormones, neurotransmitters and growth factors and 3:33 they bind to specific receptors on the cell’s surfaces or within the cells and 3:39 they trigger this effect. It’s like a cascade effect of a biochemical reaction 3:45 that ultimately changes the cellular behavior. So basically, it’s changing 3:50 the way the body’s cell structure acts. And this is why peptides can be so 3:56 incredibly powerful and therapeutic when you introduce the right peptide signal. 4:02 Now, you could theoretically redirect cellular processes toward healing, 4:07 towards metabolism, immune balance, tissue repair. Any of those things can 4:14 be manipulated to do a certain thing once we add the peptide. The challenge 4:19 in peptide medicine though lies in distinguishing between those peptides that have been rigorously studied, 4:26 proven safe and effective and approved by regulatory bodies like the FDA versus 4:31 those that exist in what we call the gray zone of a promising clinical data. 4:36 But they really lack human validation so far. And this distinction is critical because the presence of a plausible 4:43 mechanism does not guarantee safety or efficacy in living humans. So, this is 4:50 really important and we’re going to dive in and look at some of the research on all of these different peptides that are 4:56 available and I’m excited to say there’s some amazing peptides being studied right now that unfortunately are not 5:01 available. But I can’t wait to see them hit the market for us because it is going to be a gamecher as far as health 5:09 and longevity. So there is a quality control issue and there is a hidden 5:14 variable in peptide medicine with this and it’s one of the most underappreciated aspects of peptide 5:21 therapy particularly for non-FDA approved peptides. It’s quality control. 5:26 When we discuss pharmaceutical medicines, we take for granted that the pill contains what the label says. Not 5:32 always true depending on where it comes from. You guys, if you’ve heard my episodes before talk about how many of our medications are made in China and 5:41 have been contaminated with other things, you will realize that that is not always true. So, just because it has 5:48 the FDA stamp of approval on the medication, it still does not necessarily mean it’s safe and we still 5:54 need to do our homework on it. So, sorry for digressing on you guys, but you know, when we get a medication, we we 6:00 think that what the amount says is what is there, doesn’t have contaminants, it’s manufactured with good 6:06 manufacturing practices. You’ll see that listed as GMP on the bottle, and it’s been stored properly, it’s been 6:12 maintained stable, and with research peptides and compounded formulations, 6:17 none of this can be assumed. So, I will share a story with you. There was a gentleman that was purchasing these 6:24 peptides online from a research facility and um did not know that they were 6:30 coming from China and he was ordering a particular growth hormone peptide and 6:35 after a little while he had he had done fine for the few first few bottles. After a little while he started having 6:42 some complications. He started getting really irritable and angry and ragy and 6:47 he didn’t quite know what was going on. And so he decided to go get some testing done. He had some blood testing done and 6:53 his testosterone level was over 5,000. So for those of you who know what testosterone level should be for a guy, 7:00 they really shouldn’t be any higher than about 1,00200 would be absolute max that we’d want to see. Now he was taking 7:06 testosterone but not to that degree. And prior to adding this peptide, his 7:12 testosterone was very stable. What they ended up finding out was the peptide that he was getting, whoever was 7:18 manufacturing it added testosterone to the peptide. They felt like if if it had growth hormone, that was great, but if 7:25 it had growth hormone and tes testosterone, all the better. And he didn’t know that. And this is the 7:31 problem that we can have with peptides if you don’t source them properly. if you’re not working with somebody that 7:37 knows how to source them and can prove that they are what they say they are. Um, I’m sure there’s a whole bunch of 7:42 studies out there too of people getting these peptides and paying hundreds of thousands of dollars for them over their 7:48 lifetime and finding out they were nothing more than just sterile water. So, you really do need to be careful 7:53 with your quality control. Now, this kind of leads us right into the next topic that we’re going to talk about and that’s the manufacturing question, 8:00 right? The FDA approved peptides are manufactured in facilities subject to 8:05 the FDA inspection rules following our GMP regulations and these facilities 8:11 must validate their manufacturing process, demonstrate consistency batch to batch, test for purity and potency. 8:18 They need to test for bacterial endotoxins and sterility and they need to maintain detailed records. So, when a 8:25 pharmaceutical company submits a drug application, the FDA inspects the manufacturing facility as part of the 8:32 approval process. If you’re getting peptides from a different country, none of that is happening. And there are some 8:38 ways for us to determine if that is what you’re getting. Typically, the rule of thumb is if your peptides are coming 8:44 with a different colored top, every one of them has a different colored top. Those are typically being sourced out of 8:49 China. I wouldn’t say that’s 100% but that’s kind of the rule of thumb that people follow. So compoundingies these 8:56 are thearmacies that make our bio identical hormones. They can make medications in any dose or strength or 9:02 route. There are thousands of them in every not that not in every state but 9:08 there are thousands of them around the country right now. So these compoundingies are registered as 503A 9:15 facilities. They do traditional compounding for individual prescriptions, right? Like they can make 9:20 thyroid, they can make LDN, they can make estrogen. You can also have a 503b 9:27 facility, which is an outsourcing facility. And these companies produce larger batches of products. They’re they 9:34 have some oversight, but they’re less stringent than for FDA approved 9:40 manufacturers. And state boards of pharmacy regulate a 503A pharmacy. And 9:45 the FDA can inspect the 503b facility, but doesn’t preapprove any of their 9:52 compounding products. So, they can inspect it, but they don’t approve them. So, research chemicals and these 9:58 suppliers operate essentially with no oversight. They explicitly market products for research use only, not for 10:06 human consumption to avoid FDA regulation. If they put that on their 10:12 product, they don’t have to comply to what the FDA is saying. And there is no required manufacturing strategies or 10:19 standards, no required testing, no required sterility assurance, and no enforcement mechanisms if products are 10:26 mislabeled or contaminated. So basically, they don’t have the liability, but that doesn’t mean that 10:31 all of them are badies or bad suppliers. It just means they don’t have to comply 10:37 to the FDA rules. Now, there are many of these companies that I’ve seen and I’ve talked to that do do a lot of this. They 10:44 do test their product for sterility. They do test their product to make sure it is what it says it is. They don’t 10:51 have to, but they do. So, if you’re going to decide to use a company that 10:56 has research only, not for human consumption, at least ask for their 11:02 proof of testing so that you know that the product you’re getting is what it says it is and that it’s clean. Because 11:08 this is where we run into the problem is in purity. So in purity peptide 11:13 synthesis can produce not just the targeted peptide but also related 11:19 peptides with deletions, substitutions, truncations or truncations of amino 11:25 acids. Sorry. And this high performance liquid we call it uh chromatography can 11:30 separate these related impurities and quality and quantify the actual target 11:35 of the peptide content. So a certificate of analysis is what you want to ask these companies for. This shows the HPLC 11:44 the testing mechanism with greater than 95% or ideally 98% purity which 11:51 indicates a higher quality product. So this certificate of analysis can be fabricated may not represent the 11:57 specific batch being sold. It happens. We need to know not everybody is honest. Not everybody, you know, does what they 12:03 say and it does what’s right. But at least you at least they’re giving you something and you have some security. 12:10 and then choose a company that was referred to by someone else that has done some homework as well. In in 12:16 commercial research, there’s independent testing and they research peptides and this has been really shocking 12:23 variability that they’ve seen. Some products contain 50% or less of the 12:29 claimed peptide and some contained primarily degradation of the product or manufacturing impurities and some 12:36 contained bacterial endotoxins at levels that could cause fever and systemic 12:42 inflammation if it was truly injected. And I would also worry with some of those problems, you know, depending on 12:48 what impurity or bacterial endotoxin was there. If you’re using a product to boost your immune system and your immune 12:54 system is already compromised, these bacterial endotoxins can actually make you sicker instead of what you want it 13:02 to do, which is making you better. So, sterility is always an issue with anything that is manufactured, 13:08 especially things that we’re doing as an injection. Peptides are intended for injection. They must be sterile. They 13:16 must be kept safe. And pharmaceutical manufacturers conduct this sterility testing on every batch. 13:22 Compoundingarmacies should conduct sterility testing particularly for high-risisk compounded 13:28 sterile preparations and research chemical suppliers may or may not conduct any testing. So injecting 13:35 non-sterile material can cause local infections, abscesses at the injection 13:41 site and or if the bacteria enters the bloodstream could potentially be 13:46 life-threatening and you could have sepsis. Now, excuse me. We saw this 13:52 happen in a compounding pharmacy uh gosh, it’s probably been 10 years ago 13:57 now, I think. um they unfortunately had a strep uh contamination in their 14:03 product and they weren’t testing it. It was a large compounding pharmacy out of Florida and they were making products 14:08 that were being injected into the joints and um these people got very very sick 14:14 and some of them died and um some of them got very very injured by this uh 14:21 complication that happened. So it’s not like this doesn’t happen. It does, but it doesn’t happen often. And that’s what 14:28 we have to know about. And so, when we’re talking with you guys about storage and stability, it’s really 14:34 important to make sure you maintain your peptides well. So, many peptides are unstable at room temperature. They 14:41 require refrigeration or freezing. We tell everyone to make sure you’re refrigerating your peptides. That way, 14:48 there’s no question about it. when it stays cold um it prevents or slows down 14:54 the process of uh bacteria growing in it. So some of these peptides actually 14:59 degrade very rapidly in the solution and they must be reconstituted immediately before use and reconstitution of the 15:07 peptides really has limited stability often just days to weeks not months. So 15:13 improper storage, temperature, um changes during shipping or prolonged 15:19 storage of a reconstituted product can lead to degradation into inactivity or 15:25 potentially even a harmful breakdown of the product itself. So if you have a product that’s been sitting in your 15:30 refrigerator for a month or two months or 3 months or 6 months, just throw it away. It’s not going to be any good. 15:37 you’re not going to actually get the peptide and the uh potency that you’re looking for anyway out of it and the 15:44 potential of you introducing an endotoxin, a bacterial endotoxin is quite high at that point. So you just 15:50 really don’t want to take the risk, excuse me. So what practitioners, what 15:56 should we do and what should patients do? Well, for any peptide therapy, we 16:03 want to source our verification. know where the peptide product comes from. Is 16:08 it an FDA approved product? Is it a 503b compounding? A research chemical 16:14 supplier? Is there a certificate of analysis? Request and review this COA. 16:20 And you want it to show purity greater than 95% but ideally greater than 98%. 16:27 You want that identity be identity to be confirmed by mass spectromedy. Uh 16:33 sterility testing should be done. Bacterial endotoxin testing should be done. Batch number matching of the 16:39 product that you received should be done. Proper storage. You want to know that this has been refrigerated or 16:46 frozen as directed once it’s been mixed. Look at the expiration dates for reconstituting your peptides. Track that 16:53 reconstitution date and discarded accordingly like we just talked about. Monitor for your adverse effects. Even 17:01 with the perfect quality control, monitoring for adverse effects is essential with questionable quality and 17:08 vigilance is really critical here. I know it’s frustrating for a lot of patients when they have to get several 17:15 bottles and they only last a week or two. right here, you guys. This is why 17:21 they only last a short period of time because once they’re mixed, they start 17:26 to degrade and they won’t be good and you won’t get the benefit from it. So, 17:31 it’s really important with these research peptides specifically, practitioners should recognize that all 17:38 recommending products without quality assurance violates the fundamental medical principle of first do no harm. 17:45 If a patient is determined to use research peptides despite counseling, providing guidance on quality 17:52 verification, requesting those COAs, using pharmaceutical grade sources when available, proper testing, this all 17:59 reduces harm, but doesn’t constitute necessarily that recommendation. Now, 18:06 that being said, today it’s very difficult to find peptides by the compoundingies because of what the FDA 18:13 has done. So most of the peptides that are available to us have been labeled 18:18 not for human consumption, not because they’re not good products, but because 18:25 of what the FDA did. And this is how these companies have been able to 18:31 continue to provide peptides to the medical community. And if you know you 18:36 have a good company, then you’re, you know, you’re still taking the risk, right? But at the end of the day, the 18:42 reason they’re doing that is to protect themselves from the FDA, from liability. Um, so just kind of know that there is 18:50 some talk in the community with um Bobby Kennedy that this is going to change and 18:55 they are going to bring peptides back to the compounding pharmacies. Now, we don’t know which ones they’re going to 19:01 bring back. Uh, will it be all of them? Will it just be some of them? What’s going to happen here? Um, is it going to 19:07 go to the pharmaceutical companies like our GLP1s did? We don’t know what that’s going to look like quite yet. Um, but it 19:14 is coming and that is positive news. So, let’s talk now about FDA approved 19:21 peptide medications. So, this is the metabolic revolution, right? GLP1 19:28 and our dual increeting agonists. This is an exciting time. GLP-1s are amazing. 19:35 Um, a lot of people are skeptical, a lot of people love them, a lot of people hate them. Whichever side of the fence 19:42 that you’re on, I understand. But I want to talk about the science of it today 19:48 and what it actually means for people. So, the story of GLP1 glucagon like 19:54 peptide one represents one of the most significant advances in metabolic 19:59 medicine in the past several decades. GLP-1 is an accretin hormone. It’s 20:05 gutder derived peptide that potentiates insulin secretion in response to food 20:11 intake. And the body naturally produces GLP-1 in the intestinal L cells, but it 20:17 rapidly degraded by the enzyme DPP4 giving it a halflife of only about 2 20:24 minutes. So this rapid breakdown made in therapeutically impractical until 20:31 research was developed and modified the analoges that resist the enzyme degradation. So for those people who 20:39 never feel full when they’re eating, never feel satisfied when they’re done, this is because their body is either not 20:46 producing enough GLP1 or it’s not getting the signal right. And this is a 20:51 leptin issue. This is an insulin issue. It’s a GLP-1 issue. It’s a complicated 20:56 issue. This is not anything that the person is doing wrong. It’s what is happening to their body. And so GLP1s 21:03 have really revolutionized this. So one particular GLP-1 that we have is 21:09 semiglutide. And this GLP-1 agonist is what changed everything in the world of 21:16 metabolic medicine. Semiglutide is marketed as ompic for type 2 diabetes 21:23 and it’s marketed as WGOI for chronic weight management. It is a modified 21:29 GLP-1 analog with 95 or sorry 94% amino acid sequence uh homology to human 21:37 GLP-1. So it means that it’s it’s just like our own GLP-1 that we make. This 21:42 modification includes specific amino acid substitutions and the addition of C18 21:50 a fatty acid chain which allows the peptide to bind to albumin. Now this 21:56 albumin binding dramatically extends the half-life to approximately one week 22:01 enabling one weekly dosing which is a major advantage over the earlier GLP-1 22:07 agonists that require daily or twice daily injections. The mechanism by which 22:13 semiglutide works is multiaceted. At the pancreatin level, it binds to GLP-1 22:20 receptors on the pancreatic beta cells enhancing glucose depending sorry 22:27 enhancing glucose dependent insulin secretion. This glucose dependency is 22:33 crucial. It means the peptide only stimulates insulin release when blood glucose is elevated. This dramatically 22:41 reduces the hypoglycemic risk compared to insulin or even uh sulfuras. 22:47 Simultaneously semiglutide suppresses glucagon secretion from pancreatic alpha 22:53 cells further improving glycemic control. This is really amazing because 23:00 over the years when we’ve used insulin, which is also a peptide by the way, you 23:05 had to dose it just right because if you didn’t, you would produce so much insulin that it would crash the blood 23:12 sugar and then somebody would have too low of a blood sugar. They’d be hypoglycemic and they’d have to eat more 23:18 sugar and then they’d have to modify the insulin again and the person would be going up and down, up and down, up and 23:24 down all day long. And that created a lot of problems for people and so this 23:30 helps to stabilize that so it is not such an intense change. Now in the GI 23:36 tract semiglutide delays the gastric emptying particularly pronounced during 23:41 the initial weeks of therapy. This slowing of the gastric emptying contributes to the sensation of being 23:48 full and early satiety that patients often describe. However, this effect 23:54 tends to attend to weight over time as the body adapts through the appetite 24:00 suppressing effects generally persist through central mechanisms. So, when we 24:05 talk about what is actually happening, we’re slowing that digestive process down. That’s why people aren’t so 24:11 hungry. It’s why they’re not eating so much. This is why people can develop constipation with these products because 24:17 it’s slowing the body’s digestive tract down. Now some people will call this 24:22 gastroparesis. Um gastroparesis is actually different. 24:28 It is when we lose control over what’s happening in the in the colon like the 24:34 nerves and things like that just stop working. I have never seen that with the GLP1s that we prescribe in micro doing. 24:42 um it’s been documented. It can happen, but again it a lot of it is dosing and a 24:48 lot of it is staying on top of your client and what’s happening and what’s going on and what you’re doing and making sure that they do have good 24:54 motility still. So a lot of these things can be mitigated if you have problems 24:59 with them. Now one of the most profound effects of semiglutide occur in the 25:05 central nervous system. GLP-1 receptors are widely distributed in the brain 25:10 particularly in the hypothalamus and the brain stem area where we are involved in 25:15 appetite regulation. So when when wilding and colleagues published their 25:20 landmark step one trial in the New England Journal of Medicine in 2021, 25:25 they demonstrated that participants receiving 2.4 4 milligrams of semiglutide weekly achieved an average 25:32 weight loss of 14.9% of their body weight over 68 weeks. Now, I want you 25:39 guys to really understand this. We’re talking roughly 15% body weight loss 25:45 over a year, longer than a year. 52 weeks is a year, right? This is 68 25:50 weeks. So, it took longer for them to lose. We’re not talking about giving 25:55 somebody a dose to lose 15% of their body mass in a month or two. That that 26:01 is not healthy for any of us. That is not what we’re talking about doing here. Now, they compared this to placebo and 26:08 the placebo was only 2.4%. So, that is a significant difference. 26:14 And even beyond the numbers, patients reported something very qualitatively different, a reduction in what’s now 26:21 called food noise. Everybody knows what food noise is. We’ve talked about this long before GLP1. It’s that craving. 26:28 It’s that part of your brain that just keeps thinking about I want to eat something. You know, that was actually 26:34 reduced and they didn’t expect to see that happen. Now, this refers to the constant mental preoccupation with food, 26:42 the intrusive thoughts about eating, the difficulty in feeling satisfied. Semi-glutide appears to appears to 26:49 modulate reward pathways in the misolyic system reducing hedonic eating and food 26:57 cravings. Now there are also great cardiovascular effects of semiglutide 27:02 that extend beyond weight loss. Uh the sustained six and select trials 27:07 demonstrated significant reductions in major adverse cardiovascular events uh 27:14 mace in high-risisk populations. The select trial published in 2023 showed 27:20 that semiglutide reduced cardiovascular death, non-fatal myioardial inffection 27:25 and non-fatal stroke by 20% in adults with overweight or obesity and 27:31 established cardiovascular disease but without diabetes. So this suggests that 27:37 mechanisms beyond glucose control and weight loss possibly including 27:42 anti-inflammatory effects, improvements in endothelial function and favorable 27:47 changes to lipid profiles. Now I will tell you the clients that I work with that are on GLP1, 27:53 they will tell you that their inflammation has been significantly reduced. We are also seeing really 28:00 amazing results in lipid profiles. um part of its weight loss, but there is a 28:06 component to this that is lowering the triglyceride levels because it’s related to sugar and how the body’s processing 28:11 it. And we’re seeing better profiles, less need for statins as a result of 28:17 that. If if you want to listen to my episode on statins, I have one on that. Uh they are not my favorite medication. 28:24 I think it’s overprescribed and overused um and not really affecting or 28:29 addressing the problem. So these things can really be helpful. There’s also some 28:34 uh ramblings going on with GLP-1s saying that they may be able to help with 28:40 addiction in the future because of where they’re finding it affecting the brain and how it affects the food noise and 28:47 the cravings that we have for food and the addiction for food. Could it potentially help with other addictions 28:53 down the road? We’ll have to wait and see on that one. So semiglutide’s FDA prescribing information also includes a 29:00 box uh boxed warning about thyroid sea cell tumors. So in rodent studies 29:06 semiglutide caused dose dependent and treatment duration dependent sea cell 29:12 tumors at clinically relevant exposures. So while it’s unknown whether or not 29:17 semiglutide causes uh thyroid cancer tumors in humans and the rodent thyroid biology 29:26 differs significantly from humans, the drug is contraindicated in patients with a personal or family history of 29:33 medillary thyroid carcinoma or in patients with multiple endocrine neopl neoplasia syndrome type two. it is 29:42 uh contraindicated for safety effects with that. Um I have seen endocrinologists okay GLP1s to be used 29:50 in patients who’ve had other forms of thyroid cancer just not the meillary 29:55 thyroid cancer. So there is possibility there. Now the most common side effects 30:00 are gastrointestinal. It’s nausea affects about 20 to 44% of patients 30:06 depending on the formulation with diarrhea, vomiting, constipation, abdominal pain, and also frequently 30:13 reported in clinical trials. I see this in my clinic, too, especially dose dependent. Um, and it happens early on 30:20 when you’re first starting the medication, but seems to settle out over time. The one that I would add to this 30:26 that I don’t think they have on here is an increase in acid reflux. We also see that quite often uh especially in people 30:33 who suffer with acid reflux to begin with. Now these effects are typically most 30:40 pronounced during the escalation and they like I said often improve over time 30:45 but more serious but less common adverse effects include acute pancreatitis. 30:51 The medication needs to be discontinued immediately if this is confirmed. You can see some diabetic retinopathy 30:57 complications in patients with pre-existing retinopathy and acute kidney injury. Um, this usually happens 31:05 secondarily to dehydration from the GI effects. There are some gallbladder disease um that can occur and people who 31:13 have a sensitive gallbladder will describe uh discomfort with that. I’ve 31:18 even seen some people who’ve had their gallbladder out on GLP1s at the higher doses complain of similar pain that they 31:25 used to have when their gallbladder was in. So, really important to just kind of monitor these symptoms and work closely 31:32 with somebody that understands them and can be on top of them quite quickly if this happens. Excuse me. From an 31:39 integrative medicine perspective, semiglutide really represents a powerful tool, but it’s not a standalone 31:46 solution. Remember, the medication addresses one aspect of the metabolic dysfunction, the signaling systems 31:53 controlling appetite and glucose homeostasis, but it doesn’t address the root cause that led to the metabolic 32:00 disease in the first place. Patients who rely solely on the medication without addressing the ultrarocessed food 32:07 consumption, the ccadian disruptions, the chronic stress, the sleep apnea, or 32:12 underlying hormonal imbalances often experience weight regain when the medication is discontinued. 32:20 The drug is also not a substitute for addressing the emotional and psychological drivers of eating 32:26 behavior, including the unresolved trauma that may manifest as emotional eating. I think this is really important 32:33 because we don’t address the trauma issue enough with clients and we need to 32:38 be looking at that. There is a huge trauma effect out there these days that is I don’t want to say leading to or 32:45 causing but it is definitely contributing to chronic illness and it’s not being talked about enough. So we 32:52 really need to be talking about this and addressing this trauma aspect. Now the next GLP that one that I want to talk 32:59 about is trespathide. This is a dual agonist. It takes center stage. It is my 33:05 favorite GLP one. Trisepatide is marketed as Mangjaro for type 2 diabetes 33:11 and Zepbound for chronic weight management and it represents the next 33:16 evolution in increantbased therapy. This is a dual agonist a 39 amino acid 33:23 synthetic peptide structurally based on the human glucose dependent insulin tropic peptide so GIP sequence but 33:31 modified to activate both the GIP receptors and the GLP1 receptors. So the 33:37 addition of the GI GIP agonism to the GLP1 agonism appears to create this 33:46 synergistic effect that goes beyond simply adding the two mechanisms together. So the GIP like GLP-1 is an 33:55 increant hormone secreted by what is called the K cells in response to nutrient intake. It enhances glucose 34:02 dependent insulin secretion but it also effects on atapost tissue metabolism 34:09 potentially improving the insulin sensitivity in fat cells and influencing 34:14 how the body stores and metabolizes fat. So some research suggests that GIP may 34:20 also have effects on energy expenditure though this remains an area of 34:26 investigation. So basically what we’re saying is this drug may actually help 34:32 people who are insulin resistant or insulin sensitive, not just somebody who 34:38 has problems with glucose control. So, this is super exciting because it opens 34:43 up the door for all of these people for decades that we’ve been trying to manage with insulin resistance and trying to 34:50 prevent diabetes and honestly most of the time have been unsuccessful 34:56 unless you can keep your diet at 50 grams of carbs or less a day, which is extremely difficult. Um, and take some 35:04 supplements that may or may not work and or take some metformin that may or may not help. this drug actually really 35:11opens that up and helps in that capacity. So there was a clinical trial 35:17 called the surmount clinical trial which demonstrated that trespathide produces 35:22 even more substantial weight loss than semiglutide. In the surerount one trial published by uh J tree I might have said 35:31 that wrong. I apologize if I slaughtered your name and colleagues in the New York England Journal of Medicine in 2022. 35:38 Participants receiving the highest dose of trespide, which is 15 milligrams, achieved an average weight loss of 20.9% 35:47 of their body weight over 72 weeks, compared to 3.1% with placebo. This 35:54 level of weight loss approaches what’s typically only seen in beriatric surgery. So, this is amazing because if 36:02 this medication works and we don’t have to do beriatric surgery, stomach stapling basically, um, oh my gosh, it’s 36:11 amazing. There are so many complications and risks that go with stomach stapling and the different procedures that they 36:17 do these days. People don’t absorb their nutrients properly. They have to do liquid nutrients. It’s very complicated. 36:24 It’s very challenging. Many of these people gain their weight back. Um, and 36:30 this procedure is not fun to go through. So, if we could change that and change 36:35 the lives of people who’ve really been struggling, it is amazing. And I will tell you that I have seen this work. I 36:42 have seen people lose 100 150 pounds on these medications over a year or two 36:50 period of time. It is definitely slower than beriatric surgery on some standpoints, but that is okay. You don’t 36:56 want that rapid weight loss. It’s not good for you. It’s not healthy for you. It doesn’t look well. You know, we want 37:03 to do this safely and effectively in the best way that we can possibly do that for you. Now, the adverse effect profile 37:10 is similar to semiglutide. It’s dominated by gastrointestinal effects. 37:15 Nausea, diarrhea, decreased appetite, vomiting, constipation. These were all commonly reported in the surmount 37:22 trials. And like semiglutide, tricepide carries a blackbox warning regarding the 37:27 thyroid sea cell tumors based on the rodent data and it shares the same contra indications in patients with a 37:34 family history of thyroid cancer and men too. So the mechanism behind why 37:40 tepatide often produces more substantial weight loss than GLP-1. The agonism 37:45 alone remains under investigation, but it may relate to the complimentary effects on the different aspects of 37:51 energy homeostasis or to GIP’s effects on atapost tissue and potentially on 37:58 central central nervous system pathways that GLP1 alone doesn’t fully address. 38:03 Now patients often report even more profound reductions in food noise with tricepide compared to GLP1 and uh sorry 38:12 GLP1 the agonists through this is anecdotal and hasn’t been regularly 38:17 quantified in quality studies. So I’ve done both uh personally and in my 38:22 practice. I really like trespide better than semiglutide. For me I had too many side effects with semiglutide. uh I had 38:30 less side effects with trespathide. I also plateaued on semiglutide which I 38:35 didn’t really care for. And with Tresepide, I haven’t plateaued and I’ve been able 38:42 to lose about 25 pounds in um a year and a half and I’ve been able to maintain 38:49 that. Um and I continued to use it because I do have a strong family history of cardiovascular disease. And 38:56 if this could help me so that I don’t follow my family lineage with cardiovascular disease, I am all for 39:03 trying to do that. I’ve watched too many of my family members suffer from this. I’ve lost my dad at a very young age. I 39:09 lost my grandfather at a young age to it. All of their brothers to this. And I don’t want to be that same person. So 39:16 that is why I chose to do that. And I think it’s really important for us to take a look at that and understand that. 39:24 Now, I know this has been a really long podcast and I don’t typically do podcasts this long. I have a whole host 39:31 of information on additional peptides. So, I’m going to break this up for you 39:36 guys and I’m going to do another episode and we’re going to pick up where we left off here with these peptides so that we 39:43 can actually start to dive into different peptides as well. So, check 39:48 out my next podcast show when we’re going to dive into the peptides that 39:54 talk about sexual wellness, immune function, and all the other cool things 39:59 that we can do with peptides. So until then, remember to like, share, and 40:04 subscribe. It really helps us get out to other people and share our information, 40:10 and join us for our next episode as we continue the talk about peptides. 40:15 Welcome to Let’s Talk Wellness Now, where we bring expert insights directly to you. Please note that the views and 40:21 information shared by our guests are their own and do not necessarily reflect those of Let’s Talk Wellness Now, its 40:28 management, or our partners. Each affiliate, sponsor, and partner is an 40:34 independent entity with its own perspectives. Today’s content is provided forformational and educational 40:40 purposes only and should not be considered specific advice, whether financial, medical, or legal. While we 40:48 strive to present accurate and useful information, we cannot guarantee its completeness or relevance to your unique 40:56 circumstances. We encourage you to consult with a qualified professional to address your 41:01 individual needs. Your use of information from this broadcast is entirely at your own risk. By continuing 41:08 to listen, you agree to indemnify and hold Let’s Talk Wellness Now and its 41:14 associates harmless from any claims or damages arising from the use of this 41:20 content. We may update this disclaimer at any time and changes will take effect 41:26 immediately upon posting or broadcast. Thank you for tuning in. We hope you 41:31 find this episode both insightful and thought-provoking. Listener discretion 41:36 is advised.The post Episode 256 – How Peptides Work, Benefits, and FDA-Approved vs Off-Label Use Explained first appeared on Let's Talk Wellness Now.
Alex tracks down another White Whale as Surly Darkness Batch 1 lands on the show and finds itself in the mix with Holy Mountain Brewing Company's Summoning Spell, Bissell Brothers' Brewing Super Swish, and Fiddlehead Brewing's Second Fiddle. In the Beer News, a beloved SoCal brewery closes shop, BrewDog announces they're being sold but not without tons of drama, and Budweiser takes the top spot in this year's Super Bowl commercial rankings according to USA Today. To get involved with the "Life" International Barleywine Collab, click the link for info about the recipe, BSG discount, and links to help raise awareness of colon cancer. If you'd like to make a direct donation to help support Alex, head over to his GoFundMe. For more info about colon cancer and to help support the fight against it check out the Colon Cancer Foundation. Head to our Patreon for weekly exclusive content. Get the Malt Couture Officially Licensed T-shirt. Follow DontDrinkBeer on Instagram and Twitter
TwistedWave has dropped updates that feel like they were built for anyone smashing through a pile of auditions or VO edits every day. George walks through what's new in 32.3 and 32.4, including the ability to apply a Batch directly to your currently open file and map that Batch to a single keystroke. That means you can normalise to a target level, then apply a processing stack, all in one hit, without the old multi step workflow. Then there's the new Repair tool, a quick "auto heal" style fix for little clicks, mouth noises, and short waveform anomalies. Select the problem area, hit a key, and TwistedWave smooths it out. No fiddly sample drawing required. Also included, a friendly rant about software developers changing things that don't need changing, and why options matter. Sponsors, Tribooth and Austrian Audio, Making Passion Heard. Recorded using Source Connect. Edited by Andrew Peters. Mixed by Robbo.
Shandy is back with their world-famous recaps! Today they're covering episodes 10-11, aka Batch 3 of Season 10 of Netflix's Love Is Blind!Thank you to our sponsors...- Go to https://oliveandjune.com/SHANDY for 20% off your first system!- Go to https://www.functionhealth.com/SHANDY and use gift code SHANDY25 for a $25 credit towards your membership!- Go to https://laundrysauce.com and use code SHANDY for 20% off!- Go to https://www.squarespace.com/SHANDY and use code SHANDY for 10% off your first website or domain!- Go to https://www.factormeals.com/SHANDY50OFF and use code SHANDY50OFF for 50% off your first box plus free breakfast for one year!- Go to https://cozyearth.com and use code SHANDY for up to 20% off!Time Stamps:0:00 - Housekeeping0:33 - The Bowling Party, Cont'd24:54 - Christine & Vic30:58 - Amber & Jordan37:54 - Brittany & Devonta54:54 - Ashley & Alex1:20:34 - Bri & Connor1:33:53 - Emma & Mike1:46:07 - Who We Would Go ForIf you have a relationship question, write us at: dearshandy@gmail.comSubscribe and watch the episodes on YouTube! https://bit.ly/SubscribeDearShandyFollow us!Dear Shandy - https://www.instagram.com/dearshandySharleen Joynt - https://www.instagram.com/sharleenjoyntAndy Levine - https://www.instagram.com/machinelevineSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Topics Discussed:
Team lists for Vegas have dropped and SuperCoaches have been handed a gift from the fantasy Gods. With Trey Mooney, Hamish Stewart, Heilum Luki and Jermaine McEwen all named to start in the forwards, Supercoaches have been presented with arguably the best batch of mid-range buys ever. Join Tom Sangster and Wilson Smith for the lowdown.All the latest SuperCoach news and articles: linktr.ee/supercoachnrl Chapters: 1:20 - Trey Mooney 1:55 - Fletcher Sharpe 4:05 - Jacob Kiraz 5:00 - Jermaine McKewen 5:40 - Hamish Stewart 6:48 - Heilum Luki 8:53 - Knights VS Cowboys 15:36 - Bulldogs VS Dragons 21:25 - Post-Vegas QuestionsHosts: Tom Sangster: @TomSangsterSC /XWilson Smith: @wilson_smith93 /XProduced by Jack CrawleyRecorded 5pm Tuesday Feb 24, 2026See omnystudio.com/listener for privacy information.
Lyle and TJ react to the scorching opener for Michael Arroyo and a few others in the first couple Mariners Spring Training games (2:30). They then discuss the storylines they are looking for down in Arizona when they arrive (26:46).For ad-free episodes, check out our Patreon: patreon.com/marinelayerpodMerchandise, event schedule, and more: marinelayerpod.comEmail us: marinelayerpod@gmail.comCheck out Just Baseball: Click hereFollow the show on Twitter: @marinelayerpodFind us on YouTube: Click hereFind us on TikTok: https://www.tiktok.com/@marinelayerpodFind us on all Podcast Platforms: https://linktr.ee/MarineLayerPodFollow TJ on Twitter: @tjmathewsonFollow Lyle on Twitter: @lyle_goldsteinOur Sponsors:* Check out BetterHelp: https://www.betterhelp.comAdvertising Inquiries: https://redcircle.com/brandsPrivacy & Opt-Out: https://redcircle.com/privacy
As a full-time startup founder managing two brands, I create 100+ pieces of content monthly—95% by myself. In this video, I'm sharing my exact content system that helps me stay consistent without burning out.
After a brief hiatus, the boys are back!With Midjourney v8 expected next week, Drew and Rory zoom out and ask the bigger question: does v8 even matter as much as we think?Because while everyone waits for v8, Kling 3.0 and Seedance 2.0 are raising the bar, Claude Code and Claude Cowork are quietly changing how builders operate, and Claude Agents are turning workflows into autonomous systems.Meanwhile, Higgsfield is melting down in public, Hollywood is panick-maxxing, and creators are realizing that building “skills” inside LLMs might matter more than generating prettier images.This episode breaks down:• Why Midjourney v8's native 2K and edit models matter• Why personalization could be the real differentiator• How Claude Code is quietly enabling operator-level leverage• Why skill-building beats agent hype• What Kling 3.0 and Seedance 2.0 signal about video AI• The real lesson behind Higgsfield's fallout• Why the creative skill gap is widening right nowThis episode moves from Midjourney roadmap analysis to AI workflow engineering to business survival strategy.If you care about Midjourney v8, Claude Agents, Kling 3.0, Seedance 2.0, system prompts, autonomous workflows, or where creative leverage is actually going…This one isn't optional.---⏱️ Midjourney Fast Hour00:00 – Winter chaos & NYC survival03:59 – AI's quantum leap moment06:51 – Radio vs podcasts analogy08:55 – AI series vs Hollywood model10:59 – Game of Thrones AI sequel14:14 – CGI patchwork & filmmaking16:14 – AI replacing exec decisions18:03 – Seedance & model hype19:48 – Midjourney v8 timeline20:15 – Rating party (Round 2 + beyond)23:18 – 2K native resolution talk24:26 – Batch-four replacement25:48 – Edit model improvements26:14 – v8 text rendering progress27:03 – Arbitrary resolution support29:06 – Personalization in v830:36 – Mood boards as leverage32:18 – AI overwhelm & X fatigue34:12 – Claude agents & automation36:04 – “Something is happening”39:45 – AI skill-building strategy45:47 – Pattern matching workflows48:54 – Silicon Valley middle-out50:24 – Claude comedy experiment53:24 – Word clouds & ad thinking55:14 – Hollywood recycling IP58:41 – Marketing narrative engine01:03:57 – Higgsfield controversy01:11:36 – Final thoughts & sign-off
We have been sweet on donuts for centuries. While the exact origin of the sweet treat remains a mystery, many believe its invention came about in the 1600s, slowly cementing its place in history in the centuries that followed. Today, donuts are having a big moment in Duluth, thanks to a delectable new donut shop in Canal Park. Located along Lake Avenue on the main floor of the DeWitt-Seitz building, and fit with a convenient outdoor ordering and pickup window, Voyageur Donuts will have you convinced you need to eat more HOLE foods, with a varied menu of staples, signatures and seasonal specials and of course, their much buzzed about Hot Dish Donut. All pair perfectly with their carefully curated menu of coffee creations. If you haven't caught on by now, Voyageur Donuts has elevated the pastry in a pretty perfect way, inspiring Northlanders everywhere to paddle on the sweet side of life one donut at a time. Co-owner and Chief Operating Officer Tracy Owens joins us to talk all about the unique journey that led her to Duluth's most saccharine new spot: Voyageur Donuts.
WHAT YOU NEED TO KNOW?Paid Purchase Order: Management confirms the 50 kg fumed silica order is paid, with material produced and shipment logistics underway.Pilot Plant Function: The facility is performing its intended role — demonstrating scalable material production rather than prioritizing immediate revenue generation.Application Objectives: Management indicates that internal work and independent laboratory testing support that the material meets the goals for the intended application.Due Diligence Relevance: The batch is framed as a meaningful component of the technical due diligence process tied to a potential joint venture.Operational Data: Pilot plant runs are now informing more detailed assumptions, including practical considerations such as shifts, staffing, and location-dependent cost factors.Market Signaling: Management notes that milestones such as paid production runs may influence how other parties evaluate ongoing discussions.When a pilot plant progresses from demonstrating production capability to fulfilling a paid purchase order, the discussion naturally shifts from technical feasibility to real operating performance. HPQ Silicon management confirms the company has received a purchase order for 50 kilograms of fumed silica, has produced the material, and is now finalizing shipment logistics as the counterparty determines where the batch will be sent. Management explicitly states the order is paid, while underscoring an important distinction for investors: pilot plants are designed to validate commercial-scale production and generate operating data, not serve as near-term profit centers. The batch is described as part of the technical due diligence process associated with a potential joint venture, with management noting that successful material production is a necessary condition for advancing discussions. Internal testing and independent laboratory testing are described as supporting that the material meets the objectives required for the intended application.STRATEGIC IMPLICATIONSManagement emphasizes that pilot plants are not structured as profit-driven operations. Their purpose is to demonstrate that commercially valuable material can be produced and to provide the data required for designing larger-scale facilities. The discussion highlights that once systems are functioning, producing a single larger batch becomes more operationally efficient than multiple small runs. Management also indicates that a significant portion of current activity is concentrated on the joint venture process, describing both HPQ Silicon and its technical partner as heavily engaged in technical evaluation, operational analysis, and commercial discussions.INVESTOR TAKEAWAYThe significance of the paid 50 kg batch is primarily technical and strategic rather than financial. The milestone reflects pilot plant validation, supports customer-side application testing, and contributes to the refinement of detailed operating assumptions required for potential commercial expansion. As described by management, the project remains positioned within an active due-diligence phase rather than a finalized commercial rollout.
Shandy is back with their world-famous recaps! Today they're covering episodes 7-9, aka Batch 2 of Season 10 of Netflix's Love Is Blind!Thank you to our sponsors...- Go to https://www.warbyparker.com/SHANDY for 15% off 2 or more pairs of prescription glasses!- Go to https://perelelhealth.com and use code SHANDY to get 20% off your first order!- Go to https://masterclass.com/SHANDY for 15% off!- Go to https://www.quince.com/shandy for free shipping on your order and 365-day returns!- Go to https://shopremi.com/SHANDY and use code SHANDY for 50% off your custom night guard!- Go to https://cozyearth.com and use code SHANDY for up to 20% off!Time Stamps:0:00 - Housekeeping2:03 - Christine & Vic9:34 - Amber & Jordan12:57 - Ashley & Alex43:30 - Brittany & Devonta52:05 - Emma & Mike1:03:47 - Jessica & Chris1:19:21 - Bri & Connor1:24:07 - Bowling Group Party1:40:53 - Who We Would Go ForIf you have a relationship question, write us at: dearshandy@gmail.comSubscribe and watch the episodes on YouTube! https://bit.ly/SubscribeDearShandyFollow us!Dear Shandy - https://www.instagram.com/dearshandySharleen Joynt - https://www.instagram.com/sharleenjoyntAndy Levine - https://www.instagram.com/machinelevineSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Happy Tuesday! On today's show we find out how and more so why Intern John catfished Erick, we do a fresh Batch of Johns Little Secrets, We do an all NEW War of The Rose, plus we find out what your code name is. All that and more with Intern John and Your Morning Show! Make sure to also keep up to date with ALL of our podcasts we do below that have new episodes every week: The Thought Shower Let's Get Weird Crisis on Infinite Podcasts See omnystudio.com/listener for privacy information.
We have reached the season finale! In this episode of Dark Side Divas we are discussing the Star Wars - The Bad Batch episode "Kamino Lost" (s1e15). We have entered the "disaster movie" moment with this episode. There's also a lot of interesting dynamics between Crosshair and the rest of the family! Listen to hear what Stef and Chris have to say about the episode!
Jonathan Mast (the White Beard AI Guy) breaks down how video professionals can use AI to handle the administrative grunt work that's eating up their time—without sacrificing quality or creativity. He shares his proven strategies for creating consistent content, automating client onboarding, and using AI as a creative partner instead of a replacement. If you've been overwhelmed by AI or skeptical about how it fits into your business, this conversation will change your perspective. Key Takeaways Start with operations, not creativity – Use AI to handle scheduling, client communications, and admin tasks before trying to use it for creative work. These "boring" tasks are where you'll see immediate time savings. The Q&A content strategy – Record simple videos answering common client questions. You already know the answers, so there's no need for scripts or elaborate setups. Batch record them in an hour and you've got weeks of content. AI as your creative partner – Think of AI tools as amplifiers of your creativity, not replacements. Let them handle the tedious parts of editing, graphics, and client onboarding so you can focus on what you do best. Don't chase every new tool – Check in every 60-90 days on what's new in AI for video, but don't jump on every shiny object. Focus on tools that integrate with what you're already using. About Jonathan Mast Jonathan Mast stands at the forefront of AI prompting mastery, empowering businesses and entrepreneurs to leverage artificial intelligence for measurable growth. Since 1995, he has blended marketing expertise with cutting-edge technology, and over the past few years has emerged as a leading authority on practical AI implementation. With an engaged audience of nearly 600,000 AI enthusiasts and entrepreneurs (450,000+ in his active Facebook group and 100,000+ email subscribers), Jonathan is a trusted voice making complex AI concepts approachable and immediately applicable. His Perfect Prompting Framework teaches businesses how to effectively communicate with AI tools like ChatGPT, Claude, and Gemini to achieve exceptional results. As the founder of White Beard Strategies, Jonathan focuses on helping businesses and their teams leverage AI to save time, increase profits, and deliver more value to their audiences. His philosophy emphasizes AI as a tool that amplifies skill and experience rather than replacing human creativity and judgment. Jonathan's dynamic speaking style breaks down complex AI concepts into actionable steps that audiences can implement immediately. His international speaking engagements across North America, Asia, and Australia are packed with practical takeaways, and his 100+ podcast appearances demonstrate his ability to connect with and educate diverse audiences. In This Episode [00:00] Welcome to the show! [04:37] Meet Jonathan Mast [05:42] AI Marketing [17:27] AI Note Taking [21:53] Wispr Flow [26:29] Answering Questions Through AI [29:33] Why Posting Matters [30:57] Video Made Easy [44:22] Saving Time In Your Business [49:21] Connect with Jonathan [51:17] Outro Quotes "AI gives you time. It gives you space. It gives you margin and that margin lets you be the creative that you truly want to be." – Jonathan Mast "We're literally in that stage of technology right now where if you're in a business and you're not using AI, I really believe you're going to find yourself in a very bad spot within 18 months." – Jonathan Mast "If you were going to start using AI, start with your operations, start with the things you just wish somebody else would do." – Jonathan Mast "You could literally take one wasted hour a week and turn it into answering some simple questions that you already know the answers to." – Jonathan Mast "Let AI amplify your creativity. You guys have such amazing creativity but sometimes we get so bogged down in the day-to-day running our business that we can't be the creative we want." – Jonathan Mast Guest Links Connect with Jonathan Mast - https://jonathanmast.com/linktree Search "White Beard AI" to find Jonathan's content across platforms Links Find out more about the Studio Sherpas Mastermind Join the Grow Your Video Business Facebook Group Follow Ryan Koral on Instagram Follow Grow Your Video Business on Instagram Join the Studio Sherpas newsletter
Whiskey Review: Limestone Farms Heritage Family Collection Select Batch Straight Bourbon Whiskey Topic of Discussion: How we recognize a true disciple Follow us on all your podcast platforms and: Instagram: @manhoodneat X: Manhood Neat (@ManhoodNeat) / X Youtube: Manhood, Neat Podcast - YouTube Reach out: manhood.neat@gmail.com Show Notes: A Disciple Abides in Jesus' Word: John 8:31-32 - So Jesus said to the Jews who had believed him, “If you abide in my word, you are truly my disciples, and you will know the truth, and the truth will set you free.” Jesus distinguishes between those who “believed” and those who are “truly my disciples.” Abiding is the difference. Abiding means: Remaining Continuing Staying under the authority of Christ's teaching Discipleship is not validated by enthusiasm but by endurance. Modern christianity often equates belief with discipleship. Jesus equates perseverance with discipleship. A Disciple Loves Other Believers: John 13:34-35 - A new command I give you: Love one another. As I have loved you, so you must love one another. By this everyone will know that you are my disciples, if you love one another.” Love is the visible badge of discipleship Not doctrinal precision Not spiritual gifting Not ministry productivity Not generic kindness —Christ-shaped, selfless love. Christianity is communal by design. Discipleship cannot mature in isolation Where love is thin, discipleship is shallow. A disciple cannot be growing in Christ while harboring contempt toward Christ's people. A Disciple Bears Spiritual Fruit: John 15:8 - “This is to my Father's glory, that you bear much fruit, showing yourselves to be my disciples.” Fruit reveal discipleship Character transformation Obedient living Reproducing faith in others Discipleship is not static The issue is not perfection, but direction Is there increasing resemblance to Christ over time? A Disciple Places Christ Above All: Luke 14:23-33 - Large crowds were traveling with Jesus, and turning to them he said: “If anyone comes to me and does not hate father and mother, wife and children, brothers and sisters—yes, even their own life—such a person cannot be my disciple. And whoever does not carry their cross and follow me cannot be my disciple. “Suppose one of you wants to build a tower. Won't you first sit down and estimate the cost to see if you have enough money to complete it? For if you lay the foundation and are not able to finish it, everyone who sees it will ridicule you, saying, ‘This person began to build and wasn't able to finish.'” Or suppose a king is about to go to war against another king. Won't he first sit down and consider whether he is able with ten thousand men to oppose the one coming against him with twenty thousand? If he is not able, he will send a delegation while the other is still a long way off and will ask for terms of peace. In the same way, those of you who do not give up everything you have cannot be my disciples.” Jesus confronts divided allegiance To follow Christ is to reorder every competing loyalty Family, comfort, ambition, self-rule….none outrank Him. This is not an emotional rejection of others Its supreme allegiance to Christ Half-hearted discipleship is self-deception Is He ultimate or supplemental? What we protect most reveals what we worship most
In this episode of The Bourbon Life Podcast, Mark and Matt welcome Barry and Tori Brinegar into The Bourbon Life Studios for a conversation that's equal parts bourbon history, industry insight, and good-natured chaos. Barry is well known across the whiskey world as the former Co-Founder and National Brand Ambassador of RD1 Spirits, and he's joined by his wife Tori—affectionately known around here as the First Lady of Bourbon. Across three segments, the guys dig into Barry's journey through the bourbon industry, the founding and growth of RD1, and what life looks like after stepping away from a brand he helped build from the ground up. Barry shares behind-the-scenes stories from the road as a brand ambassador, lessons learned building a bourbon company, and a few tales that probably shouldn't be repeated—but thankfully were anyway. Tori jumps in with her perspective on life in the bourbon world, keeping Barry grounded, and what it's really like being married to a guy whose job revolves around whiskey. As always, the pours are flowing and the reviews are honest. This episode features tastings of: Henry McKenna 10 Year Bottled in Bond Stagg Jr. Batch 12 Jack Daniel's Single Barrel Barrel Proof Rye There's plenty of laughter, a few strong opinions, and a whole lot of great bourbon conversation in this one. Pull up a chair, pour yourself something neat, and enjoy a fun, candid sit-down with one of bourbon's most recognizable personalities—and the woman who keeps him in line. This Episode is sponsored by District 7 and The Kitchen Table at the James B. Beam Distilling Co.
Returning to their old way of ranking beers, Alex's video game and Stephen's wrestling rating systems are back for this batch's motley lineup of beers. Bell's Brewery's Hopslam finally makes it on the show. A throwback to the IBU Wars, Three Floyds Arctic Panzer Wolf, takes the Malty Boyz™ back to the mid-2000s. Alex's leather jacket-gate inspires a massive collaboration with Barreled Souls Grifter and Friends strong ale. Then a Drekker's smoothie sour has Alex struggling to wrap his head around... well, craft beer. In the Beer News, Midnight Sun shutters an iconic taproom location while a bear poop fueled press release makes the rounds through the press. To get involved with the "Life" International Barleywine Collab, click the link for info about the recipe, BSG discount, and links to help raise awareness of colon cancer. If you'd like to make a direct donation to help support Alex, head over to his GoFundMe. For more info about colon cancer and to help support the fight against it check out the Colon Cancer Foundation. Head to our Patreon for weekly exclusive content. Get the Malt Couture Officially Licensed T-shirt. Follow DontDrinkBeer on Instagram and Twitter
From rewriting Google's search stack in the early 2000s to reviving sparse trillion-parameter models and co-designing TPUs with frontier ML research, Jeff Dean has quietly shaped nearly every layer of the modern AI stack. As Chief AI Scientist at Google and a driving force behind Gemini, Jeff has lived through multiple scaling revolutions from CPUs and sharded indices to multimodal models that reason across text, video, and code.Jeff joins us to unpack what it really means to “own the Pareto frontier,” why distillation is the engine behind every Flash model breakthrough, how energy (in picojoules) not FLOPs is becoming the true bottleneck, what it was like leading the charge to unify all of Google's AI teams, and why the next leap won't come from bigger context windows alone, but from systems that give the illusion of attending to trillions of tokens.We discuss:* Jeff's early neural net thesis in 1990: parallel training before it was cool, why he believed scaling would win decades early, and the “bigger model, more data, better results” mantra that held for 15 years* The evolution of Google Search: sharding, moving the entire index into memory in 2001, softening query semantics pre-LLMs, and why retrieval pipelines already resemble modern LLM systems* Pareto frontier strategy: why you need both frontier “Pro” models and low-latency “Flash” models, and how distillation lets smaller models surpass prior generations* Distillation deep dive: ensembles → compression → logits as soft supervision, and why you need the biggest model to make the smallest one good* Latency as a first-class objective: why 10–50x lower latency changes UX entirely, and how future reasoning workloads will demand 10,000 tokens/sec* Energy-based thinking: picojoules per bit, why moving data costs 1000x more than a multiply, batching through the lens of energy, and speculative decoding as amortization* TPU co-design: predicting ML workloads 2–6 years out, speculative hardware features, precision reduction, sparsity, and the constant feedback loop between model architecture and silicon* Sparse models and “outrageously large” networks: trillions of parameters with 1–5% activation, and why sparsity was always the right abstraction* Unified vs. specialized models: abandoning symbolic systems, why general multimodal models tend to dominate vertical silos, and when vertical fine-tuning still makes sense* Long context and the illusion of scale: beyond needle-in-a-haystack benchmarks toward systems that narrow trillions of tokens to 117 relevant documents* Personalized AI: attending to your emails, photos, and documents (with permission), and why retrieval + reasoning will unlock deeply personal assistants* Coding agents: 50 AI interns, crisp specifications as a new core skill, and how ultra-low latency will reshape human–agent collaboration* Why ideas still matter: transformers, sparsity, RL, hardware, systems — scaling wasn't blind; the pieces had to multiply togetherShow Notes:* Gemma 3 Paper* Gemma 3* Gemini 2.5 Report* Jeff Dean's “Software Engineering Advice fromBuilding Large-Scale Distributed Systems” Presentation (with Back of the Envelope Calculations)* Latency Numbers Every Programmer Should Know by Jeff Dean* The Jeff Dean Facts* Jeff Dean Google Bio* Jeff Dean on “Important AI Trends” @Stanford AI Club* Jeff Dean & Noam Shazeer — 25 years at Google (Dwarkesh)—Jeff Dean* LinkedIn: https://www.linkedin.com/in/jeff-dean-8b212555* X: https://x.com/jeffdeanGoogle* https://google.com* https://deepmind.googleFull Video EpisodeTimestamps00:00:04 — Introduction: Alessio & Swyx welcome Jeff Dean, chief AI scientist at Google, to the Latent Space podcast00:00:30 — Owning the Pareto Frontier & balancing frontier vs low-latency models00:01:31 — Frontier models vs Flash models + role of distillation00:03:52 — History of distillation and its original motivation00:05:09 — Distillation's role in modern model scaling00:07:02 — Model hierarchy (Flash, Pro, Ultra) and distillation sources00:07:46 — Flash model economics & wide deployment00:08:10 — Latency importance for complex tasks00:09:19 — Saturation of some tasks and future frontier tasks00:11:26 — On benchmarks, public vs internal00:12:53 — Example long-context benchmarks & limitations00:15:01 — Long-context goals: attending to trillions of tokens00:16:26 — Realistic use cases beyond pure language00:18:04 — Multimodal reasoning and non-text modalities00:19:05 — Importance of vision & motion modalities00:20:11 — Video understanding example (extracting structured info)00:20:47 — Search ranking analogy for LLM retrieval00:23:08 — LLM representations vs keyword search00:24:06 — Early Google search evolution & in-memory index00:26:47 — Design principles for scalable systems00:28:55 — Real-time index updates & recrawl strategies00:30:06 — Classic “Latency numbers every programmer should know”00:32:09 — Cost of memory vs compute and energy emphasis00:34:33 — TPUs & hardware trade-offs for serving models00:35:57 — TPU design decisions & co-design with ML00:38:06 — Adapting model architecture to hardware00:39:50 — Alternatives: energy-based models, speculative decoding00:42:21 — Open research directions: complex workflows, RL00:44:56 — Non-verifiable RL domains & model evaluation00:46:13 — Transition away from symbolic systems toward unified LLMs00:47:59 — Unified models vs specialized ones00:50:38 — Knowledge vs reasoning & retrieval + reasoning00:52:24 — Vertical model specialization & modules00:55:21 — Token count considerations for vertical domains00:56:09 — Low resource languages & contextual learning00:59:22 — Origins: Dean's early neural network work01:10:07 — AI for coding & human–model interaction styles01:15:52 — Importance of crisp specification for coding agents01:19:23 — Prediction: personalized models & state retrieval01:22:36 — Token-per-second targets (10k+) and reasoning throughput01:23:20 — Episode conclusion and thanksTranscriptAlessio Fanelli [00:00:04]: Hey everyone, welcome to the Latent Space podcast. This is Alessio, founder of Kernel Labs, and I'm joined by Swyx, editor of Latent Space. Shawn Wang [00:00:11]: Hello, hello. We're here in the studio with Jeff Dean, chief AI scientist at Google. Welcome. Thanks for having me. It's a bit surreal to have you in the studio. I've watched so many of your talks, and obviously your career has been super legendary. So, I mean, congrats. I think the first thing must be said, congrats on owning the Pareto Frontier.Jeff Dean [00:00:30]: Thank you, thank you. Pareto Frontiers are good. It's good to be out there.Shawn Wang [00:00:34]: Yeah, I mean, I think it's a combination of both. You have to own the Pareto Frontier. You have to have like frontier capability, but also efficiency, and then offer that range of models that people like to use. And, you know, some part of this was started because of your hardware work. Some part of that is your model work, and I'm sure there's lots of secret sauce that you guys have worked on cumulatively. But, like, it's really impressive to see it all come together in, like, this slittily advanced.Jeff Dean [00:01:04]: Yeah, yeah. I mean, I think, as you say, it's not just one thing. It's like a whole bunch of things up and down the stack. And, you know, all of those really combine to help make UNOS able to make highly capable large models, as well as, you know, software techniques to get those large model capabilities into much smaller, lighter weight models that are, you know, much more cost effective and lower latency, but still, you know, quite capable for their size. Yeah.Alessio Fanelli [00:01:31]: How much pressure do you have on, like, having the lower bound of the Pareto Frontier, too? I think, like, the new labs are always trying to push the top performance frontier because they need to raise more money and all of that. And you guys have billions of users. And I think initially when you worked on the CPU, you were thinking about, you know, if everybody that used Google, we use the voice model for, like, three minutes a day, they were like, you need to double your CPU number. Like, what's that discussion today at Google? Like, how do you prioritize frontier versus, like, we have to do this? How do we actually need to deploy it if we build it?Jeff Dean [00:02:03]: Yeah, I mean, I think we always want to have models that are at the frontier or pushing the frontier because I think that's where you see what capabilities now exist that didn't exist at the sort of slightly less capable last year's version or last six months ago version. At the same time, you know, we know those are going to be really useful for a bunch of use cases, but they're going to be a bit slower and a bit more expensive than people might like for a bunch of other broader models. So I think what we want to do is always have kind of a highly capable sort of affordable model that enables a whole bunch of, you know, lower latency use cases. People can use them for agentic coding much more readily and then have the high-end, you know, frontier model that is really useful for, you know, deep reasoning, you know, solving really complicated math problems, those kinds of things. And it's not that. One or the other is useful. They're both useful. So I think we'd like to do both. And also, you know, through distillation, which is a key technique for making the smaller models more capable, you know, you have to have the frontier model in order to then distill it into your smaller model. So it's not like an either or choice. You sort of need that in order to actually get a highly capable, more modest size model. Yeah.Alessio Fanelli [00:03:24]: I mean, you and Jeffrey came up with the solution in 2014.Jeff Dean [00:03:28]: Don't forget, L'Oreal Vinyls as well. Yeah, yeah.Alessio Fanelli [00:03:30]: A long time ago. But like, I'm curious how you think about the cycle of these ideas, even like, you know, sparse models and, you know, how do you reevaluate them? How do you think about in the next generation of model, what is worth revisiting? Like, yeah, they're just kind of like, you know, you worked on so many ideas that end up being influential, but like in the moment, they might not feel that way necessarily. Yeah.Jeff Dean [00:03:52]: I mean, I think distillation was originally motivated because we were seeing that we had a very large image data set at the time, you know, 300 million images that we could train on. And we were seeing that if you create specialists for different subsets of those image categories, you know, this one's going to be really good at sort of mammals, and this one's going to be really good at sort of indoor room scenes or whatever, and you can cluster those categories and train on an enriched stream of data after you do pre-training on a much broader set of images. You get much better performance. If you then treat that whole set of maybe 50 models you've trained as a large ensemble, but that's not a very practical thing to serve, right? So distillation really came about from the idea of, okay, what if we want to actually serve that and train all these independent sort of expert models and then squish it into something that actually fits in a form factor that you can actually serve? And that's, you know, not that different from what we're doing today. You know, often today we're instead of having an ensemble of 50 models. We're having a much larger scale model that we then distill into a much smaller scale model.Shawn Wang [00:05:09]: Yeah. A part of me also wonders if distillation also has a story with the RL revolution. So let me maybe try to articulate what I mean by that, which is you can, RL basically spikes models in a certain part of the distribution. And then you have to sort of, well, you can spike models, but usually sometimes... It might be lossy in other areas and it's kind of like an uneven technique, but you can probably distill it back and you can, I think that the sort of general dream is to be able to advance capabilities without regressing on anything else. And I think like that, that whole capability merging without loss, I feel like it's like, you know, some part of that should be a distillation process, but I can't quite articulate it. I haven't seen much papers about it.Jeff Dean [00:06:01]: Yeah, I mean, I tend to think of one of the key advantages of distillation is that you can have a much smaller model and you can have a very large, you know, training data set and you can get utility out of making many passes over that data set because you're now getting the logits from the much larger model in order to sort of coax the right behavior out of the smaller model that you wouldn't otherwise get with just the hard labels. And so, you know, I think that's what we've observed. Is you can get, you know, very close to your largest model performance with distillation approaches. And that seems to be, you know, a nice sweet spot for a lot of people because it enables us to kind of, for multiple Gemini generations now, we've been able to make the sort of flash version of the next generation as good or even substantially better than the previous generations pro. And I think we're going to keep trying to do that because that seems like a good trend to follow.Shawn Wang [00:07:02]: So, Dara asked, so it was the original map was Flash Pro and Ultra. Are you just sitting on Ultra and distilling from that? Is that like the mother load?Jeff Dean [00:07:12]: I mean, we have a lot of different kinds of models. Some are internal ones that are not necessarily meant to be released or served. Some are, you know, our pro scale model and we can distill from that as well into our Flash scale model. So I think, you know, it's an important set of capabilities to have and also inference time scaling. It can also be a useful thing to improve the capabilities of the model.Shawn Wang [00:07:35]: And yeah, yeah, cool. Yeah. And obviously, I think the economy of Flash is what led to the total dominance. I think the latest number is like 50 trillion tokens. I don't know. I mean, obviously, it's changing every day.Jeff Dean [00:07:46]: Yeah, yeah. But, you know, by market share, hopefully up.Shawn Wang [00:07:50]: No, I mean, there's no I mean, there's just the economics wise, like because Flash is so economical, like you can use it for everything. Like it's in Gmail now. It's in YouTube. Like it's yeah. It's in everything.Jeff Dean [00:08:02]: We're using it more in our search products of various AI mode reviews.Shawn Wang [00:08:05]: Oh, my God. Flash past the AI mode. Oh, my God. Yeah, that's yeah, I didn't even think about that.Jeff Dean [00:08:10]: I mean, I think one of the things that is quite nice about the Flash model is not only is it more affordable, it's also a lower latency. And I think latency is actually a pretty important characteristic for these models because we're going to want models to do much more complicated things that are going to involve, you know, generating many more tokens from when you ask the model to do so. So, you know, if you're going to ask the model to do something until it actually finishes what you ask it to do, because you're going to ask now, not just write me a for loop, but like write me a whole software package to do X or Y or Z. And so having low latency systems that can do that seems really important. And Flash is one direction, one way of doing that. You know, obviously our hardware platforms enable a bunch of interesting aspects of our, you know, serving stack as well, like TPUs, the interconnect between. Chips on the TPUs is actually quite, quite high performance and quite amenable to, for example, long context kind of attention operations, you know, having sparse models with lots of experts. These kinds of things really, really matter a lot in terms of how do you make them servable at scale.Alessio Fanelli [00:09:19]: Yeah. Does it feel like there's some breaking point for like the proto Flash distillation, kind of like one generation delayed? I almost think about almost like the capability as a. In certain tasks, like the pro model today is a saturated, some sort of task. So next generation, that same task will be saturated at the Flash price point. And I think for most of the things that people use models for at some point, the Flash model in two generation will be able to do basically everything. And how do you make it economical to like keep pushing the pro frontier when a lot of the population will be okay with the Flash model? I'm curious how you think about that.Jeff Dean [00:09:59]: I mean, I think that's true. If your distribution of what people are asking people, the models to do is stationary, right? But I think what often happens is as the models become more capable, people ask them to do more, right? So, I mean, I think this happens in my own usage. Like I used to try our models a year ago for some sort of coding task, and it was okay at some simpler things, but wouldn't do work very well for more complicated things. And since then, we've improved dramatically on the more complicated coding tasks. And now I'll ask it to do much more complicated things. And I think that's true, not just of coding, but of, you know, now, you know, can you analyze all the, you know, renewable energy deployments in the world and give me a report on solar panel deployment or whatever. That's a very complicated, you know, more complicated task than people would have asked a year ago. And so you are going to want more capable models to push the frontier in the absence of what people ask the models to do. And that also then gives us. Insight into, okay, where does the, where do things break down? How can we improve the model in these, these particular areas, uh, in order to sort of, um, make the next generation even better.Alessio Fanelli [00:11:11]: Yeah. Are there any benchmarks or like test sets they use internally? Because it's almost like the same benchmarks get reported every time. And it's like, all right, it's like 99 instead of 97. Like, how do you have to keep pushing the team internally to it? Or like, this is what we're building towards. Yeah.Jeff Dean [00:11:26]: I mean, I think. Benchmarks, particularly external ones that are publicly available. Have their utility, but they often kind of have a lifespan of utility where they're introduced and maybe they're quite hard for current models. You know, I, I like to think of the best kinds of benchmarks are ones where the initial scores are like 10 to 20 or 30%, maybe, but not higher. And then you can sort of work on improving that capability for, uh, whatever it is, the benchmark is trying to assess and get it up to like 80, 90%, whatever. I, I think once it hits kind of 95% or something, you get very diminishing returns from really focusing on that benchmark, cuz it's sort of, it's either the case that you've now achieved that capability, or there's also the issue of leakage in public data or very related kind of data being, being in your training data. Um, so we have a bunch of held out internal benchmarks that we really look at where we know that wasn't represented in the training data at all. There are capabilities that we want the model to have. Um, yeah. Yeah. Um, that it doesn't have now, and then we can work on, you know, assessing, you know, how do we make the model better at these kinds of things? Is it, we need different kind of data to train on that's more specialized for this particular kind of task. Do we need, um, you know, a bunch of, uh, you know, architectural improvements or some sort of, uh, model capability improvements, you know, what would help make that better?Shawn Wang [00:12:53]: Is there, is there such an example that you, uh, a benchmark inspired in architectural improvement? Like, uh, I'm just kind of. Jumping on that because you just.Jeff Dean [00:13:02]: Uh, I mean, I think some of the long context capability of the, of the Gemini models that came, I guess, first in 1.5 really were about looking at, okay, we want to have, um, you know,Shawn Wang [00:13:15]: immediately everyone jumped to like completely green charts of like, everyone had, I was like, how did everyone crack this at the same time? Right. Yeah. Yeah.Jeff Dean [00:13:23]: I mean, I think, um, and once you're set, I mean, as you say that needed single needle and a half. Hey, stack benchmark is really saturated for at least context links up to 1, 2 and K or something. Don't actually have, you know, much larger than 1, 2 and 8 K these days or two or something. We're trying to push the frontier of 1 million or 2 million context, which is good because I think there are a lot of use cases where. Yeah. You know, putting a thousand pages of text or putting, you know, multiple hour long videos and the context and then actually being able to make use of that as useful. Try to, to explore the über graduation are fairly large. But the single needle in a haystack benchmark is sort of saturated. So you really want more complicated, sort of multi-needle or more realistic, take all this content and produce this kind of answer from a long context that sort of better assesses what it is people really want to do with long context. Which is not just, you know, can you tell me the product number for this particular thing?Shawn Wang [00:14:31]: Yeah, it's retrieval. It's retrieval within machine learning. It's interesting because I think the more meta level I'm trying to operate at here is you have a benchmark. You're like, okay, I see the architectural thing I need to do in order to go fix that. But should you do it? Because sometimes that's an inductive bias, basically. It's what Jason Wei, who used to work at Google, would say. Exactly the kind of thing. Yeah, you're going to win. Short term. Longer term, I don't know if that's going to scale. You might have to undo that.Jeff Dean [00:15:01]: I mean, I like to sort of not focus on exactly what solution we're going to derive, but what capability would you want? And I think we're very convinced that, you know, long context is useful, but it's way too short today. Right? Like, I think what you would really want is, can I attend to the internet while I answer my question? Right? But that's not going to happen. I think that's going to be solved by purely scaling the existing solutions, which are quadratic. So a million tokens kind of pushes what you can do. You're not going to do that to a trillion tokens, let alone, you know, a billion tokens, let alone a trillion. But I think if you could give the illusion that you can attend to trillions of tokens, that would be amazing. You'd find all kinds of uses for that. You would have attend to the internet. You could attend to the pixels of YouTube and the sort of deeper representations that we can find. You could attend to the form for a single video, but across many videos, you know, on a personal Gemini level, you could attend to all of your personal state with your permission. So like your emails, your photos, your docs, your plane tickets you have. I think that would be really, really useful. And the question is, how do you get algorithmic improvements and system level improvements that get you to something where you actually can attend to trillions of tokens? Right. In a meaningful way. Yeah.Shawn Wang [00:16:26]: But by the way, I think I did some math and it's like, if you spoke all day, every day for eight hours a day, you only generate a maximum of like a hundred K tokens, which like very comfortably fits.Jeff Dean [00:16:38]: Right. But if you then say, okay, I want to be able to understand everything people are putting on videos.Shawn Wang [00:16:46]: Well, also, I think that the classic example is you start going beyond language into like proteins and whatever else is extremely information dense. Yeah. Yeah.Jeff Dean [00:16:55]: I mean, I think one of the things about Gemini's multimodal aspects is we've always wanted it to be multimodal from the start. And so, you know, that sometimes to people means text and images and video sort of human-like and audio, audio, human-like modalities. But I think it's also really useful to have Gemini know about non-human modalities. Yeah. Like LIDAR sensor data from. Yes. Say, Waymo vehicles or. Like robots or, you know, various kinds of health modalities, x-rays and MRIs and imaging and genomics information. And I think there's probably hundreds of modalities of data where you'd like the model to be able to at least be exposed to the fact that this is an interesting modality and has certain meaning in the world. Where even if you haven't trained on all the LIDAR data or MRI data, you could have, because maybe that's not, you know, it doesn't make sense in terms of trade-offs of. You know, what you include in your main pre-training data mix, at least including a little bit of it is actually quite useful. Yeah. Because it sort of tempts the model that this is a thing.Shawn Wang [00:18:04]: Yeah. Do you believe, I mean, since we're on this topic and something I just get to ask you all the questions I always wanted to ask, which is fantastic. Like, are there some king modalities, like modalities that supersede all the other modalities? So a simple example was Vision can, on a pixel level, encode text. And DeepSeq had this DeepSeq CR paper that did that. Vision. And Vision has also been shown to maybe incorporate audio because you can do audio spectrograms and that's, that's also like a Vision capable thing. Like, so, so maybe Vision is just the king modality and like. Yeah.Jeff Dean [00:18:36]: I mean, Vision and Motion are quite important things, right? Motion. Well, like video as opposed to static images, because I mean, there's a reason evolution has evolved eyes like 23 independent ways, because it's such a useful capability for sensing the world around you, which is really what we want these models to be. So I think the only thing that we can be able to do is interpret the things we're seeing or the things we're paying attention to and then help us in using that information to do things. Yeah.Shawn Wang [00:19:05]: I think motion, you know, I still want to shout out, I think Gemini, still the only native video understanding model that's out there. So I use it for YouTube all the time. Nice.Jeff Dean [00:19:15]: Yeah. Yeah. I mean, it's actually, I think people kind of are not necessarily aware of what the Gemini models can actually do. Yeah. Like I have an example I've used in one of my talks. It had like, it was like a YouTube highlight video of 18 memorable sports moments across the last 20 years or something. So it has like Michael Jordan hitting some jump shot at the end of the finals and, you know, some soccer goals and things like that. And you can literally just give it the video and say, can you please make me a table of what all these different events are? What when the date is when they happened? And a short description. And so you get like now an 18 row table of that information extracted from the video, which is, you know, not something most people think of as like a turn video into sequel like table.Alessio Fanelli [00:20:11]: Has there been any discussion inside of Google of like, you mentioned tending to the whole internet, right? Google, it's almost built because a human cannot tend to the whole internet and you need some sort of ranking to find what you need. Yep. That ranking is like much different for an LLM because you can expect a person to look at maybe the first five, six links in a Google search versus for an LLM. Should you expect to have 20 links that are highly relevant? Like how do you internally figure out, you know, how do we build the AI mode that is like maybe like much broader search and span versus like the more human one? Yeah.Jeff Dean [00:20:47]: I mean, I think even pre-language model based work, you know, our ranking systems would be built to start. I mean, I think even pre-language model based work, you know, our ranking systems would be built to start. With a giant number of web pages in our index, many of them are not relevant. So you identify a subset of them that are relevant with very lightweight kinds of methods. You know, you're down to like 30,000 documents or something. And then you gradually refine that to apply more and more sophisticated algorithms and more and more sophisticated sort of signals of various kinds in order to get down to ultimately what you show, which is, you know, the final 10 results or, you know, 10 results plus. Other kinds of information. And I think an LLM based system is not going to be that dissimilar, right? You're going to attend to trillions of tokens, but you're going to want to identify, you know, what are the 30,000 ish documents that are with the, you know, maybe 30 million interesting tokens. And then how do you go from that into what are the 117 documents I really should be paying attention to in order to carry out the tasks that the user has asked? And I think, you know, you can imagine systems where you have, you know, a lot of highly parallel processing to identify those initial 30,000 candidates, maybe with very lightweight kinds of models. Then you have some system that sort of helps you narrow down from 30,000 to the 117 with maybe a little bit more sophisticated model or set of models. And then maybe the final model is the thing that looks. So the 117 things that might be your most capable model. So I think it has to, it's going to be some system like that, that is really enables you to give the illusion of attending to trillions of tokens. Sort of the way Google search gives you, you know, not the illusion, but you are searching the internet, but you're finding, you know, a very small subset of things that are, that are relevant.Shawn Wang [00:22:47]: Yeah. I often tell a lot of people that are not steeped in like Google search history that, well, you know, like Bert was. Like he was like basically immediately inside of Google search and that improves results a lot, right? Like I don't, I don't have any numbers off the top of my head, but like, I'm sure you guys, that's obviously the most important numbers to Google. Yeah.Jeff Dean [00:23:08]: I mean, I think going to an LLM based representation of text and words and so on enables you to get out of the explicit hard notion of, of particular words having to be on the page, but really getting at the notion of this topic of this page or this page. Paragraph is highly relevant to this query. Yeah.Shawn Wang [00:23:28]: I don't think people understand how much LLMs have taken over all these very high traffic system, very high traffic. Yeah. Like it's Google, it's YouTube. YouTube has this like semantics ID thing where it's just like every token or every item in the vocab is a YouTube video or something that predicts the video using a code book, which is absurd to me for YouTube size.Jeff Dean [00:23:50]: And then most recently GROK also for, for XAI, which is like, yeah. I mean, I'll call out even before LLMs were used extensively in search, we put a lot of emphasis on softening the notion of what the user actually entered into the query.Shawn Wang [00:24:06]: So do you have like a history of like, what's the progression? Oh yeah.Jeff Dean [00:24:09]: I mean, I actually gave a talk in, uh, I guess, uh, web search and data mining conference in 2009, uh, where we never actually published any papers about the origins of Google search, uh, sort of, but we went through sort of four or five or six. generations, four or five or six generations of, uh, redesigning of the search and retrieval system, uh, from about 1999 through 2004 or five. And that talk is really about that evolution. And one of the things that really happened in 2001 was we were sort of working to scale the system in multiple dimensions. So one is we wanted to make our index bigger, so we could retrieve from a larger index, which always helps your quality in general. Uh, because if you don't have the page in your index, you're going to not do well. Um, and then we also needed to scale our capacity because we were, our traffic was growing quite extensively. Um, and so we had, you know, a sharded system where you have more and more shards as the index grows, you have like 30 shards. And then if you want to double the index size, you make 60 shards so that you can bound the latency by which you respond for any particular user query. Um, and then as traffic grows, you add, you add more and more replicas of each of those. And so we eventually did the math that realized that in a data center where we had say 60 shards and, um, you know, 20 copies of each shard, we now had 1200 machines, uh, with disks. And we did the math and we're like, Hey, one copy of that index would actually fit in memory across 1200 machines. So in 2001, we introduced, uh, we put our entire index in memory and what that enabled from a quality perspective was amazing. Um, and so we had more and more replicas of each of those. Before you had to be really careful about, you know, how many different terms you looked at for a query, because every one of them would involve a disk seek on every one of the 60 shards. And so you, as you make your index bigger, that becomes even more inefficient. But once you have the whole index in memory, it's totally fine to have 50 terms you throw into the query from the user's original three or four word query, because now you can add synonyms like restaurant and restaurants and cafe and, uh, you know, things like that. Uh, bistro and all these things. And you can suddenly start, uh, sort of really, uh, getting at the meaning of the word as opposed to the exact semantic form the user typed in. And that was, you know, 2001, very much pre LLM, but really it was about softening the, the strict definition of what the user typed in order to get at the meaning.Alessio Fanelli [00:26:47]: What are like principles that you use to like design the systems, especially when you have, I mean, in 2001, the internet is like. Doubling, tripling every year in size is not like, uh, you know, and I think today you kind of see that with LLMs too, where like every year the jumps in size and like capabilities are just so big. Are there just any, you know, principles that you use to like, think about this? Yeah.Jeff Dean [00:27:08]: I mean, I think, uh, you know, first, whenever you're designing a system, you want to understand what are the sort of design parameters that are going to be most important in designing that, you know? So, you know, how many queries per second do you need to handle? How big is the internet? How big is the index you need to handle? How much data do you need to keep for every document in the index? How are you going to look at it when you retrieve things? Um, what happens if traffic were to double or triple, you know, will that system work well? And I think a good design principle is you're going to want to design a system so that the most important characteristics could scale by like factors of five or 10, but probably not beyond that because often what happens is if you design a system for X. And something suddenly becomes a hundred X, that would enable a very different point in the design space that would not make sense at X. But all of a sudden at a hundred X makes total sense. So like going from a disk space index to a in memory index makes a lot of sense once you have enough traffic, because now you have enough replicas of the sort of state on disk that those machines now actually can hold, uh, you know, a full copy of the, uh, index and memory. Yeah. And that all of a sudden enabled. A completely different design that wouldn't have been practical before. Yeah. Um, so I'm, I'm a big fan of thinking through designs in your head, just kind of playing with the design space a little before you actually do a lot of writing of code. But, you know, as you said, in the early days of Google, we were growing the index, uh, quite extensively. We were growing the update rate of the index. So the update rate actually is the parameter that changed the most. Surprising. So it used to be once a month.Shawn Wang [00:28:55]: Yeah.Jeff Dean [00:28:56]: And then we went to a system that could update any particular page in like sub one minute. Okay.Shawn Wang [00:29:02]: Yeah. Because this is a competitive advantage, right?Jeff Dean [00:29:04]: Because all of a sudden news related queries, you know, if you're, if you've got last month's news index, it's not actually that useful for.Shawn Wang [00:29:11]: News is a special beast. Was there any, like you could have split it onto a separate system.Jeff Dean [00:29:15]: Well, we did. We launched a Google news product, but you also want news related queries that people type into the main index to also be sort of updated.Shawn Wang [00:29:23]: So, yeah, it's interesting. And then you have to like classify whether the page is, you have to decide which pages should be updated and what frequency. Oh yeah.Jeff Dean [00:29:30]: There's a whole like, uh, system behind the scenes that's trying to decide update rates and importance of the pages. So even if the update rate seems low, you might still want to recrawl important pages quite often because, uh, the likelihood they change might be low, but the value of having updated is high.Shawn Wang [00:29:50]: Yeah, yeah, yeah, yeah. Uh, well, you know, yeah. This, uh, you know, mention of latency and, and saving things to this reminds me of one of your classics, which I have to bring up, which is latency numbers. Every programmer should know, uh, was there a, was it just a, just a general story behind that? Did you like just write it down?Jeff Dean [00:30:06]: I mean, this has like sort of eight or 10 different kinds of metrics that are like, how long does a cache mistake? How long does branch mispredict take? How long does a reference domain memory take? How long does it take to send, you know, a packet from the U S to the Netherlands or something? Um,Shawn Wang [00:30:21]: why Netherlands, by the way, or is it, is that because of Chrome?Jeff Dean [00:30:25]: Uh, we had a data center in the Netherlands, um, so, I mean, I think this gets to the point of being able to do the back of the envelope calculations. So these are sort of the raw ingredients of those, and you can use them to say, okay, well, if I need to design a system to do image search and thumb nailing or something of the result page, you know, how, what I do that I could pre-compute the image thumbnails. I could like. Try to thumbnail them on the fly from the larger images. What would that do? How much dis bandwidth than I need? How many des seeks would I do? Um, and you can sort of actually do thought experiments in, you know, 30 seconds or a minute with the sort of, uh, basic, uh, basic numbers at your fingertips. Uh, and then as you sort of build software using higher level libraries, you kind of want to develop the same intuitions for how long does it take to, you know, look up something in this particular kind of.Shawn Wang [00:31:21]: I'll see you next time.Shawn Wang [00:31:51]: Which is a simple byte conversion. That's nothing interesting. I wonder if you have any, if you were to update your...Jeff Dean [00:31:58]: I mean, I think it's really good to think about calculations you're doing in a model, either for training or inference.Jeff Dean [00:32:09]: Often a good way to view that is how much state will you need to bring in from memory, either like on-chip SRAM or HBM from the accelerator. Attached memory or DRAM or over the network. And then how expensive is that data motion relative to the cost of, say, an actual multiply in the matrix multiply unit? And that cost is actually really, really low, right? Because it's order, depending on your precision, I think it's like sub one picodule.Shawn Wang [00:32:50]: Oh, okay. You measure it by energy. Yeah. Yeah.Jeff Dean [00:32:52]: Yeah. I mean, it's all going to be about energy and how do you make the most energy efficient system. And then moving data from the SRAM on the other side of the chip, not even off the off chip, but on the other side of the same chip can be, you know, a thousand picodules. Oh, yeah. And so all of a sudden, this is why your accelerators require batching. Because if you move, like, say, the parameter of a model from SRAM on the, on the chip into the multiplier unit, that's going to cost you a thousand picodules. So you better make use of that, that thing that you moved many, many times with. So that's where the batch dimension comes in. Because all of a sudden, you know, if you have a batch of 256 or something, that's not so bad. But if you have a batch of one, that's really not good.Shawn Wang [00:33:40]: Yeah. Yeah. Right.Jeff Dean [00:33:41]: Because then you paid a thousand picodules in order to do your one picodule multiply.Shawn Wang [00:33:46]: I have never heard an energy-based analysis of batching.Jeff Dean [00:33:50]: Yeah. I mean, that's why people batch. Yeah. Ideally, you'd like to use batch size one because the latency would be great.Shawn Wang [00:33:56]: The best latency.Jeff Dean [00:33:56]: But the energy cost and the compute cost inefficiency that you get is quite large. So, yeah.Shawn Wang [00:34:04]: Is there a similar trick like, like, like you did with, you know, putting everything in memory? Like, you know, I think obviously NVIDIA has caused a lot of waves with betting very hard on SRAM with Grok. I wonder if, like, that's something that you already saw with, with the TPUs, right? Like that, that you had to. Uh, to serve at your scale, uh, you probably sort of saw that coming. Like what, what, what hardware, uh, innovations or insights were formed because of what you're seeing there?Jeff Dean [00:34:33]: Yeah. I mean, I think, you know, TPUs have this nice, uh, sort of regular structure of 2D or 3D meshes with a bunch of chips connected. Yeah. And each one of those has HBM attached. Um, I think for serving some kinds of models, uh, you know, you, you pay a lot higher cost. Uh, and time latency, um, bringing things in from HBM than you do bringing them in from, uh, SRAM on the chip. So if you have a small enough model, you can actually do model parallelism, spread it out over lots of chips and you actually get quite good throughput improvements and latency improvements from doing that. And so you're now sort of striping your smallish scale model over say 16 or 64 chips. Uh, but as if you do that and it all fits in. In SRAM, uh, that can be a big win. So yeah, that's not a surprise, but it is a good technique.Alessio Fanelli [00:35:27]: Yeah. What about the TPU design? Like how much do you decide where the improvements have to go? So like, this is like a good example of like, is there a way to bring the thousand picojoules down to 50? Like, is it worth designing a new chip to do that? The extreme is like when people say, oh, you should burn the model on the ASIC and that's kind of like the most extreme thing. How much of it? Is it worth doing an hardware when things change so quickly? Like what was the internal discussion? Yeah.Jeff Dean [00:35:57]: I mean, we, we have a lot of interaction between say the TPU chip design architecture team and the sort of higher level modeling, uh, experts, because you really want to take advantage of being able to co-design what should future TPUs look like based on where we think the sort of ML research puck is going, uh, in some sense, because, uh, you know, as a hardware designer for ML and in particular, you're trying to design a chip starting today and that design might take two years before it even lands in a data center. And then it has to sort of be a reasonable lifetime of the chip to take you three, four or five years. So you're trying to predict two to six years out where, what ML computations will people want to run two to six years out in a very fast changing field. And so having people with interest. Interesting ML research ideas of things we think will start to work in that timeframe or will be more important in that timeframe, uh, really enables us to then get, you know, interesting hardware features put into, you know, TPU N plus two, where TPU N is what we have today.Shawn Wang [00:37:10]: Oh, the cycle time is plus two.Jeff Dean [00:37:12]: Roughly. Wow. Because, uh, I mean, sometimes you can squeeze some changes into N plus one, but, you know, bigger changes are going to require the chip. Yeah. Design be earlier in its lifetime design process. Um, so whenever we can do that, it's generally good. And sometimes you can put in speculative features that maybe won't cost you much chip area, but if it works out, it would make something, you know, 10 times as fast. And if it doesn't work out, well, you burned a little bit of tiny amount of your chip area on that thing, but it's not that big a deal. Uh, sometimes it's a very big change and we want to be pretty sure this is going to work out. So we'll do like lots of carefulness. Uh, ML experimentation to show us, uh, this is actually the, the way we want to go. Yeah.Alessio Fanelli [00:37:58]: Is there a reverse of like, we already committed to this chip design so we can not take the model architecture that way because it doesn't quite fit?Jeff Dean [00:38:06]: Yeah. I mean, you, you definitely have things where you're going to adapt what the model architecture looks like so that they're efficient on the chips that you're going to have for both training and inference of that, of that, uh, generation of model. So I think it kind of goes both ways. Um, you know, sometimes you can take advantage of, you know, lower precision things that are coming in a future generation. So you can, might train it at that lower precision, even if the current generation doesn't quite do that. Mm.Shawn Wang [00:38:40]: Yeah. How low can we go in precision?Jeff Dean [00:38:43]: Because people are saying like ternary is like, uh, yeah, I mean, I'm a big fan of very low precision because I think that gets, that saves you a tremendous amount of time. Right. Because it's picojoules per bit that you're transferring and reducing the number of bits is a really good way to, to reduce that. Um, you know, I think people have gotten a lot of luck, uh, mileage out of having very low bit precision things, but then having scaling factors that apply to a whole bunch of, uh, those, those weights. Scaling. How does it, how does it, okay.Shawn Wang [00:39:15]: Interesting. You, so low, low precision, but scaled up weights. Yeah. Huh. Yeah. Never considered that. Yeah. Interesting. Uh, w w while we're on this topic, you know, I think there's a lot of, um, uh, this, the concept of precision at all is weird when we're sampling, you know, uh, we just, at the end of this, we're going to have all these like chips that I'll do like very good math. And then we're just going to throw a random number generator at the start. So, I mean, there's a movement towards, uh, energy based, uh, models and processors. I'm just curious if you've, obviously you've thought about it, but like, what's your commentary?Jeff Dean [00:39:50]: Yeah. I mean, I think. There's a bunch of interesting trends though. Energy based models is one, you know, diffusion based models, which don't sort of sequentially decode tokens is another, um, you know, speculative decoding is a way that you can get sort of an equivalent, very small.Shawn Wang [00:40:06]: Draft.Jeff Dean [00:40:07]: Batch factor, uh, for like you predict eight tokens out and that enables you to sort of increase the effective batch size of what you're doing by a factor of eight, even, and then you maybe accept five or six of those tokens. So you get. A five, a five X improvement in the amortization of moving weights, uh, into the multipliers to do the prediction for the, the tokens. So these are all really good techniques and I think it's really good to look at them from the lens of, uh, energy, real energy, not energy based models, um, and, and also latency and throughput, right? If you look at things from that lens, that sort of guides you to. Two solutions that are gonna be, uh, you know, better from, uh, you know, being able to serve larger models or, you know, equivalent size models more cheaply and with lower latency.Shawn Wang [00:41:03]: Yeah. Well, I think, I think I, um, it's appealing intellectually, uh, haven't seen it like really hit the mainstream, but, um, I do think that, uh, there's some poetry in the sense that, uh, you know, we don't have to do, uh, a lot of shenanigans if like we fundamentally. Design it into the hardware. Yeah, yeah.Jeff Dean [00:41:23]: I mean, I think there's still a, there's also sort of the more exotic things like analog based, uh, uh, computing substrates as opposed to digital ones. Uh, I'm, you know, I think those are super interesting cause they can be potentially low power. Uh, but I think you often end up wanting to interface that with digital systems and you end up losing a lot of the power advantages in the digital to analog and analog to digital conversions. You end up doing, uh, at the sort of boundaries. And periphery of that system. Um, I still think there's a tremendous distance we can go from where we are today in terms of energy efficiency with sort of, uh, much better and specialized hardware for the models we care about.Shawn Wang [00:42:05]: Yeah.Alessio Fanelli [00:42:06]: Um, any other interesting research ideas that you've seen, or like maybe things that you cannot pursue a Google that you would be interested in seeing researchers take a step at, I guess you have a lot of researchers. Yeah, I guess you have enough, but our, our research.Jeff Dean [00:42:21]: Our research portfolio is pretty broad. I would say, um, I mean, I think, uh, in terms of research directions, there's a whole bunch of, uh, you know, open problems and how do you make these models reliable and able to do much longer, kind of, uh, more complex tasks that have lots of subtasks. How do you orchestrate, you know, maybe one model that's using other models as tools in order to sort of build, uh, things that can accomplish, uh, you know, much more. Yeah. Significant pieces of work, uh, collectively, then you would ask a single model to do. Um, so that's super interesting. How do you get more verifiable, uh, you know, how do you get RL to work for non-verifiable domains? I think it's a pretty interesting open problem because I think that would broaden out the capabilities of the models, the improvements that you're seeing in both math and coding. Uh, if we could apply those to other less verifiable domains, because we've come up with RL techniques that actually enable us to do that. Uh, effectively, that would, that would really make the models improve quite a lot. I think.Alessio Fanelli [00:43:26]: I'm curious, like when we had Noam Brown on the podcast, he said, um, they already proved you can do it with deep research. Um, you kind of have it with AI mode in a way it's not verifiable. I'm curious if there's any thread that you think is interesting there. Like what is it? Both are like information retrieval of JSON. So I wonder if it's like the retrieval is like the verifiable part. That you can score or what are like, yeah, yeah. How, how would you model that, that problem?Jeff Dean [00:43:55]: Yeah. I mean, I think there are ways of having other models that can evaluate the results of what a first model did, maybe even retrieving. Can you have another model that says, is this things, are these things you retrieved relevant? Or can you rate these 2000 things you retrieved to assess which ones are the 50 most relevant or something? Um, I think those kinds of techniques are actually quite effective. Sometimes I can even be the same model, just prompted differently to be a, you know, a critic as opposed to a, uh, actual retrieval system. Yeah.Shawn Wang [00:44:28]: Um, I do think like there, there is that, that weird cliff where like, it feels like we've done the easy stuff and then now it's, but it always feels like that every year. It's like, oh, like we know, we know, and the next part is super hard and nobody's figured it out. And, uh, exactly with this RLVR thing where like everyone's talking about, well, okay, how do we. the next stage of the non-verifiable stuff. And everyone's like, I don't know, you know, Ellen judge.Jeff Dean [00:44:56]: I mean, I feel like the nice thing about this field is there's lots and lots of smart people thinking about creative solutions to some of the problems that we all see. Uh, because I think everyone sort of sees that the models, you know, are great at some things and they fall down around the edges of those things and, and are not as capable as we'd like in those areas. And then coming up with good techniques and trying those. And seeing which ones actually make a difference is sort of what the whole research aspect of this field is, is pushing forward. And I think that's why it's super interesting. You know, if you think about two years ago, we were struggling with GSM, eight K problems, right? Like, you know, Fred has two rabbits. He gets three more rabbits. How many rabbits does he have? That's a pretty far cry from the kinds of mathematics that the models can, and now you're doing IMO and Erdos problems in pure language. Yeah. Yeah. Pure language. So that is a really, really amazing jump in capabilities in, you know, in a year and a half or something. And I think, um, for other areas, it'd be great if we could make that kind of leap. Uh, and you know, we don't exactly see how to do it for some, some areas, but we do see it for some other areas and we're going to work hard on making that better. Yeah.Shawn Wang [00:46:13]: Yeah.Alessio Fanelli [00:46:14]: Like YouTube thumbnail generation. That would be very helpful. We need that. That would be AGI. We need that.Shawn Wang [00:46:20]: That would be. As far as content creators go.Jeff Dean [00:46:22]: I guess I'm not a YouTube creator, so I don't care that much about that problem, but I guess, uh, many people do.Shawn Wang [00:46:27]: It does. Yeah. It doesn't, it doesn't matter. People do judge books by their covers as it turns out. Um, uh, just to draw a bit on the IMO goal. Um, I'm still not over the fact that a year ago we had alpha proof and alpha geometry and all those things. And then this year we were like, screw that we'll just chuck it into Gemini. Yeah. What's your reflection? Like, I think this, this question about. Like the merger of like symbolic systems and like, and, and LMS, uh, was a very much core belief. And then somewhere along the line, people would just said, Nope, we'll just all do it in the LLM.Jeff Dean [00:47:02]: Yeah. I mean, I think it makes a lot of sense to me because, you know, humans manipulate symbols, but we probably don't have like a symbolic representation in our heads. Right. We have some distributed representation that is neural net, like in some way of lots of different neurons. And activation patterns firing when we see certain things and that enables us to reason and plan and, you know, do chains of thought and, you know, roll them back now that, that approach for solving the problem doesn't seem like it's going to work. I'm going to try this one. And, you know, in a lot of ways we're emulating what we intuitively think, uh, is happening inside real brains in neural net based models. So it never made sense to me to have like completely separate. Uh, discrete, uh, symbolic things, and then a completely different way of, of, uh, you know, thinking about those things.Shawn Wang [00:47:59]: Interesting. Yeah. Uh, I mean, it's maybe seems obvious to you, but it wasn't obvious to me a year ago. Yeah.Jeff Dean [00:48:06]: I mean, I do think like that IMO with, you know, translating to lean and using lean and then the next year and also a specialized geometry model. And then this year switching to a single unified model. That is roughly the production model with a little bit more inference budget, uh, is actually, you know, quite good because it shows you that the capabilities of that general model have improved dramatically and, and now you don't need the specialized model. This is actually sort of very similar to the 2013 to 16 era of machine learning, right? Like it used to be, people would train separate models for lots of different, each different problem, right? I have, I want to recognize street signs and something. So I train a street sign. Recognition recognition model, or I want to, you know, decode speech recognition. I have a speech model, right? I think now the era of unified models that do everything is really upon us. And the question is how well do those models generalize to new things they've never been asked to do and they're getting better and better.Shawn Wang [00:49:10]: And you don't need domain experts. Like one of my, uh, so I interviewed ETA who was on, who was on that team. Uh, and he was like, yeah, I, I don't know how they work. I don't know where the IMO competition was held. I don't know the rules of it. I just trained the models, the training models. Yeah. Yeah. And it's kind of interesting that like people with these, this like universal skill set of just like machine learning, you just give them data and give them enough compute and they can kind of tackle any task, which is the bitter lesson, I guess. I don't know. Yeah.Jeff Dean [00:49:39]: I mean, I think, uh, general models, uh, will win out over specialized ones in most cases.Shawn Wang [00:49:45]: Uh, so I want to push there a bit. I think there's one hole here, which is like, uh. There's this concept of like, uh, maybe capacity of a model, like abstractly a model can only contain the number of bits that it has. And, uh, and so it, you know, God knows like Gemini pro is like one to 10 trillion parameters. We don't know, but, uh, the Gemma models, for example, right? Like a lot of people want like the open source local models that are like that, that, that, and, and, uh, they have some knowledge, which is not necessary, right? Like they can't know everything like, like you have the. The luxury of you have the big model and big model should be able to capable of everything. But like when, when you're distilling and you're going down to the small models, you know, you're actually memorizing things that are not useful. Yeah. And so like, how do we, I guess, do we want to extract that? Can we, can we divorce knowledge from reasoning, you know?Jeff Dean [00:50:38]: Yeah. I mean, I think you do want the model to be most effective at reasoning if it can retrieve things, right? Because having the model devote precious parameter space. To remembering obscure facts that could be looked up is actually not the best use of that parameter space, right? Like you might prefer something that is more generally useful in more settings than this obscure fact that it has. Um, so I think that's always attention at the same time. You also don't want your model to be kind of completely detached from, you know, knowing stuff about the world, right? Like it's probably useful to know how long the golden gate be. Bridges just as a general sense of like how long are bridges, right? And, uh, it should have that kind of knowledge. It maybe doesn't need to know how long some teeny little bridge in some other more obscure part of the world is, but, uh, it does help it to have a fair bit of world knowledge and the bigger your model is, the more you can have. Uh, but I do think combining retrieval with sort of reasoning and making the model really good at doing multiple stages of retrieval. Yeah.Shawn Wang [00:51:49]: And reasoning through the intermediate retrieval results is going to be a, a pretty effective way of making the model seem much more capable, because if you think about, say, a personal Gemini, yeah, right?Jeff Dean [00:52:01]: Like we're not going to train Gemini on my email. Probably we'd rather have a single model that, uh, we can then use and use being able to retrieve from my email as a tool and have the model reason about it and retrieve from my photos or whatever, uh, and then make use of that and have multiple. Um, you know, uh, stages of interaction. that makes sense.Alessio Fanelli [00:52:24]: Do you think the vertical models are like, uh, interesting pursuit? Like when people are like, oh, we're building the best healthcare LLM, we're building the best law LLM, are those kind of like short-term stopgaps or?Jeff Dean [00:52:37]: No, I mean, I think, I think vertical models are interesting. Like you want them to start from a pretty good base model, but then you can sort of, uh, sort of viewing them, view them as enriching the data. Data distribution for that particular vertical domain for healthcare, say, um, we're probably not going to train or for say robotics. We're probably not going to train Gemini on all possible robotics data. We, you could train it on because we want it to have a balanced set of capabilities. Um, so we'll expose it to some robotics data, but if you're trying to build a really, really good robotics model, you're going to want to start with that and then train it on more robotics data. And then maybe that would. It's multilingual translation capability, but improve its robotics capabilities. And we're always making these kind of, uh, you know, trade-offs in the data mix that we train the base Gemini models on. You know, we'd love to include data from 200 more languages and as much data as we have for those languages, but that's going to displace some other capabilities of the model. It won't be as good at, um, you know, Pearl programming, you know, it'll still be good at Python programming. Cause we'll include it. Enough. Of that, but there's other long tail computer languages or coding capabilities that it may suffer on or multi, uh, multimodal reasoning capabilities may suffer. Cause we didn't get to expose it to as much data there, but it's really good at multilingual things. So I, I think some combination of specialized models, maybe more modular models. So it'd be nice to have the capability to have those 200 languages, plus this awesome robotics model, plus this awesome healthcare, uh, module that all can be knitted together to work in concert and called upon in different circumstances. Right? Like if I have a health related thing, then it should enable using this health module in conjunction with the main base model to be even better at those kinds of things. Yeah.Shawn Wang [00:54:36]: Installable knowledge. Yeah.Jeff Dean [00:54:37]: Right.Shawn Wang [00:54:38]: Just download as a, as a package.Jeff Dean [00:54:39]: And some of that installable stuff can come from retrieval, but some of it probably should come from preloaded training on, you know, uh, a hundred billion tokens or a trillion tokens of health data. Yeah.Shawn Wang [00:54:51]: And for listeners, I think, uh, I will highlight the Gemma three end paper where they, there was a little bit of that, I think. Yeah.Alessio Fanelli [00:54:56]: Yeah. I guess the question is like, how many billions of tokens do you need to outpace the frontier model improvements? You know, it's like, if I have to make this model better healthcare and the main. Gemini model is still improving. Do I need 50 billion tokens? Can I do it with a hundred, if I need a trillion healthcare tokens, it's like, they're probably not out there that you don't have, you know, I think that's really like the.Jeff Dean [00:55:21]: Well, I mean, I think healthcare is a particularly challenging domain, so there's a lot of healthcare data that, you know, we don't have access to appropriately, but there's a lot of, you know, uh, healthcare organizations that want to train models on their own data. That is not public healthcare data, uh, not public health. But public healthcare data. Um, so I think there are opportunities there to say, partner with a large healthcare organization and train models for their use that are going to be, you know, more bespoke, but probably, uh, might be better than a general model trained on say, public data. Yeah.Shawn Wang [00:55:58]: Yeah. I, I believe, uh, by the way, also this is like somewhat related to the language conversation. Uh, I think one of your, your favorite examples was you can put a low resource language in the context and it just learns. Yeah.Jeff Dean [00:56:09]: Oh, yeah, I think the example we used was Calamon, which is truly low resource because it's only spoken by, I think 120 people in the world and there's no written text.Shawn Wang [00:56:20]: So, yeah. So you can just do it that way. Just put it in the context. Yeah. Yeah. But I think your whole data set in the context, right.Jeff Dean [00:56:27]: If you, if you take a language like, uh, you know, Somali or something, there is a fair bit of Somali text in the world that, uh, or Ethiopian Amharic or something, um, you know, we probably. Yeah. Are not putting all the data from those languages into the Gemini based training. We put some of it, but if you put more of it, you'll improve the capabilities of those models.Shawn Wang [00:56:49]: Yeah.Jeff Dean [00:56:49]:
Shandy is back with their world-famous recaps! Today they're covering episodes 1-6, aka Batch 1 of Season 10 of Netflix's Love Is Blind!Thank you to our sponsors...- Go to https://revolve.com/SHANDY and use code SHANDY for 15% off your first order!- Go to https://rula.com/shandy and take the first step towards improved mental health!- Go to https://ollie.com/SHANDY and use code SHANDY for 60% off your first box!- Go to https://oneskin.co/SHANDY and use code SHANDY for 15% off!- Go to https://piquelife.com/SHANDY for 20% off your order!- Go to https://shopremi.com/SHANDY and use CODE SHANDY for 50% off your custom night guard!Time Stamps:0:00 - Housekeeping4:02 - Christine & Vic13:09 - Amber & Jordan17:20 - Ashely & Alex35:27 - Bri & Conner (& Chris)53:57 - Chris & Jessica1:02:31 - Emma & Mike (& Conner & Steven)1:26:56- Kevan & Keya & Tyler1:44:33 - Brittany & Devonta1:51:37 - Who We Would Go ForIf you have a relationship question, write us at: dearshandy@gmail.comSubscribe and watch the episodes on YouTube! https://bit.ly/SubscribeDearShandyFollow us!Dear Shandy - https://www.instagram.com/dearshandySharleen Joynt - https://www.instagram.com/sharleenjoyntAndy Levine - https://www.instagram.com/machinelevineSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Film Photography Podcast Episode 367 - February 10, 2026 / Michael Raso is joined by Mat Marrash for a hands-on discussion about several new motion picture film stocks. The episode dives into black-and-white and color offerings across multiple formats, with practical insight into how these films behave, where they shine, and who they're best suited for. Batch it on YouTube - https://youtu.be/1KZiWuTN2kU?si=9FXWtM4SvMYhsdUF The conversation covers Svema r32 BW 16mm, Ferrania P30 BW in both 8mm and 16mm, Super 8 Wolfen 200D Color, and FPP Color Test Film 16mm. Michael and Mat share real-world observations, creative considerations, and thoughts on why these films matter to experimental shooters, filmmakers, and analog enthusiasts alike.
Hunter is in trouble!!!!!! In this episode of Dark Side Divas we discuss the Star Wars - The Bad Batch episode "Return to Kamino" (s1e15). Crosshair is setting up a trap for Clone Force 99, but does he want to kill his brothers? Meanwhile the Empire is making plans for Kamino. Join us to hear how Star Wars hurt us (as per usual).
It's This Week in Bourbon for February 6th 2026. The court-appointed receiver for Uncle Nearest, presents a grim financial outlook for the whiskey brand, The Kentucky Bourbon Trail adds 10 new stops, and Brown Forman is releasing King of Kentucky Small Batch.Show Notes: Uncle Nearest faces insolvency with $164M debt and revenue shortfall ABC Fine Wine & Spirits enters Colorado with Applejack acquisition Kentucky Bourbon Trail expands to record 68 stops statewide Barrell Craft Spirits consolidates blending operations to original Gilmore facility Supreme Court weighs legality of out-of-state alcohol shipping bans Kentucky Bourbon industry economic impact surges to $10.6 billion Jim Beam taps Kenan Thompson for 2026 "Refresh Your Season" campaign Yellowstone Bourbon partners with Vital Ground Foundation for grizzly conservation King of Kentucky announces 250th Anniversary Small Batch three-part series Shortbarrel launches Four Grain Straight Bourbon flagship for nationwide distribution Buzzard's Roost unveils 5-year-old Four Grain Double Oak Bourbon Chattanooga Whiskey debuts Irish-style Batch 047: Single Pot Still Learn more about your ad choices. Visit megaphone.fm/adchoices
Canucks PxP voice Brendan Batchelor joins the show. Batch talks what to watch for coming out of the Olympic break for the Canucks, the most intriguing Canucks player at the Olympics, and his favourite non-Hockey Olympic sport. Then, Dimitri Filipovic joins the show. The guys assemble the ultimate Team Canada lineup. Later, what are Canucks players going to do during this break? This podcast is produced by Dominic Sramaty and Elan CharkThe views and opinions expressed in this podcast are those of the hosts and guests and do not necessarily reflect the position of Rogers Media Inc. or any affiliate.
Shopify Masters | The ecommerce business and marketing podcast for ambitious entrepreneurs
COTTO's founder sold out her initial production run in 30 minutes by building her audience first. Use her social media techniques to validate demand before you're ready to sell. Subscribe and watch Shopify Masters on YouTube!Sign up for your FREE Shopify Trial here.
President Trump attempted earlier today to distance himself from the dead pedophile and sex trafficker Jeffrey Epstein. A look at the revelations inside the Epstein files. Plus, an urgent search for the mother of Today Show anchor Savannah Guthrie. Learn more about your ad choices. Visit podcastchoices.com/adchoices
Hey guys, Here is my complete method! Prep Plan-Room & scheduling map-Zoning & destination boxes List - get your task list Progress Nervous system regulation Breathing Somatic grounding Spatial grounding Routing objects Task list Process Clean up Take boxes to destinations Batch and schedule tasks Protect Protect your space Protect your energy Protect your capacity If you want to go deeper and have support decluttering your home consistently, the year-long program is open. You can find all the details at declutteryourchaos.com. ✨Come home to yourself. ✨ Head to Cozy Earth and use my code DECLUTTER for 20% off and experience the softest sheets you can find: https://cozyearth.com/ If this episode helped you, please leave a review or share it with someone who needs it. Looking forward to seeing your progress in the free Facebook group. To join click below... https://www.facebook.com/groups/declutteryourchaos/ Download my free decluttering planner here: https://declutteryourchaos.com/decluttering-planner Let's connect:
Message us ANONYMOUSLYOur pals the Gremlins have returned in a cash grab the likes of which you have never seen before in “Gremlins 2: The new Batch” (1990 Dir. Dante) TOYS! VIDEO GAMES! Mercccchhhhhhhandi$$$$$ing! We bring in returning guest Jessica Erin Martin to discuss sequelitis, pastiche, franchises, coincidence, cohesion, and plot holes galore amongst a cavalcade of puppet stars! It's a lively discussion, that doesn't end until Paul says so, or dies, whichever comes first. 2/3!****A member of the “Review Review,” family is in the fight of her life, you can help! - TAP/CLICK Support the show**All episodes contain explicit language**Artwork - Ben McFaddenReview Review Intro/Outro Theme - Jamie Henwood"What Are We Watching" & "Whatcha been up to?" Themes - Matthew Fosket"Fun Facts" Theme - Chris Olds/Paul RootLead-Ins Edited/Conceptualized by - Ben McFaddenProduced by - Ben McFadden & Paul RootConcept - Paul Root
Declutter Your Chaos - Minimalism, Decluttering, Home Organization
Hey guys, Here is my complete method! Prep Plan-Room & scheduling map-Zoning & destination boxes List - get your task list Progress Nervous system regulation Breathing Somatic grounding Spatial grounding Routing objects Task list Process Clean up Take boxes to destinations Batch and schedule tasks Protect Protect your space Protect your energy Protect your capacity If you want to go deeper and have support decluttering your home consistently, the year-long program is open. You can find all the details at declutteryourchaos.com. ✨Come home to yourself. ✨ Head to Cozy Earth and use my code DECLUTTER for 20% off and experience the softest sheets you can find: https://cozyearth.com/ If this episode helped you, please leave a review or share it with someone who needs it. Looking forward to seeing your progress in the free Facebook group. To join click below... https://www.facebook.com/groups/declutteryourchaos/ Download my free decluttering planner here: https://declutteryourchaos.com/decluttering-planner Let's connect:
In this episode of Dark Side Divas we discuss the Star Wars - The Bad Batch episode "War Mantle" (s1e14). Clone Force 99 gets a distress call from Rex, who is going to ask for a favor. Meanwhile we met a "TK Trooper" for the first time, and we go back to Kamino. Listen to this episode to hear what the divas have to say!
In this episode of the Roast It Yourself Podcast, we dive into a practical roasting challenge many home roasters face: scaling up batch size without wasting coffee. Stephen Burnett and Catherine Mansell respond to a listener question from Greg, a longtime DIY roaster who recently upgraded to an Aillio Bullet R2 and is looking to move from 250-gram batches to the Bullet's sweet spot of 1.5 pounds—without burning through pounds of coffee in trial and error. Catherine breaks down what really changes when scaling a roast, focusing on the critical role of preheat temperature, thermal mass, and how to maintain similar roast curves as batch size increases. Along the way, they talk Kenya AA as a use-case, share realistic expectations for profile translation, and offer guidance that applies not just to the Bullet, but to thoughtful, data-driven roasting in general. As always, the episode blends hands-on technical advice with real-world experience, making it valuable for home roasters leveling up and professionals refining their process. Got a roasting question of your own? Send it to questions@riypod.com NOTES: Follow Our Instagram Account @RIY_POD CHECK US OUT HERE: Coffee Bean Corral YouTube Coffee Bean Corral Website Current Crop Roasting Shop Website Rancher Wholesale Website
If you've ever sat down to plan and felt instantly overwhelmed, this episode is for you. ESL teachers juggle multiple grade levels, language levels, paperwork, meetings, and constant interruptions — all with the same planning time as everyone else. It's no wonder planning can feel impossible. In today's episode, we talk honestly about why ESL lesson planning feels so heavy and what you can do when time is limited but your students still deserve meaningful instruction. You'll walk away with simple, realistic strategies that help you plan faster, smarter, and with less stress — even during your busiest seasons. Before you dive in, don't forget to explore engaging, scaffolded ESL resources designed to save you time:
From January 27, 2025: In a live conversation on January 23, Lawfare Editor-in-Chief Benjamin Wittes spoke to Lawfare Senior Editors Scott R. Anderson, Anna Bower, Quinta Jurecic, and Alan Rozenshtein and assistant law professor at Pace University Amelia Wilson about the first batch of executive orders by President Trump in his second term, including suspending enforcement of the TikTok ban, the use of the military at the border, the birthright citizenship order, and the legal challenges some of these orders are facing.To receive ad-free podcasts, become a Lawfare Material Supporter at www.patreon.com/lawfare. You can also support Lawfare by making a one-time donation at https://givebutter.com/lawfare-institute.Support this show http://supporter.acast.com/lawfare. Hosted on Acast. See acast.com/privacy for more information.
Things have got worse for Andrew Mountbatten-Windsor - as fresh Epstein files show how deeply he was involved with Jeffery Epstein. Photos of Andrew crouched on all fours and touching an unidentified woman have been released. The British Prime Minister's suggested Andrew go to the U.S.senate to explain himself. Royal commentator Richard Fitzwilliams told Mike Hosking that Keir Starmer has toughened his line. He says some of Andrew's emails with Epstein occurred when he had previously claimed publicly he hadn't been in touch. LISTEN ABOVESee omnystudio.com/listener for privacy information.
Learn how to use "Developer Logic" to automate your print-on-demand business, allowing you to launch 120 product listings at once using MyDesigns and Printify. I'm sharing the exact systems that helped me transition from a senior web developer to generating multi-million dollar sales so you can spend less time on manual tasks and more time scaling your passive income.
Alex hunts down the best and most sought after whiskeys released in 2025! E.H. Taylor BTAC, Russell's Reserve 15 Year, Jack Daniel's Single Barrel Special Release Tanyard HIll Rye, Bardstown Bourbon Company Distillery Reserve Hokkaido Mizunara Oak Barrel Finish, and Hill Farmstead Whistle Pig Rye 10 Year all compete for that coveted spot atop the Malt Couture Power Rankings. In the Beer News, the TTB gets served a lawsuit to allow meads, ciders, and fruit wines to display vintages on their labels and the world's oldest monastic brewery in Germany is sold. Thanks to Amory's Tomb Brewing Co. for sponsoring this episode. Visit their newly reopened tap room in Maynard, Massachusetts. Look for them at the New England Real Ale Exhibition from March 25-28 and at Widowmaker's Hopsmokerfest in April! Follow them on IG @AmorysTomb! To get involved with the "Life" International Barleywine Collab, click the link for info about the recipe, BSG discount, and links to help raise awareness of colon cancer. If you'd like to make a direct donation to help support Alex, head over to his GoFundMe. For more info about colon cancer and to help support the fight against it check out the Colon Cancer Foundation. Head to our Patreon for weekly exclusive content. Get the Malt Couture Officially Licensed T-shirt. Follow DontDrinkBeer on Instagram and Twitter
Food FAQ - Learn How to Cook: Cooking, Kitchen Tips, and Lots of Love
Stop paying twenty dollars for a tiny bottle of "artisanal" syrup when one ten-dollar bag of dried flowers makes enough mixer for forty people!
How does gut health affect MS – and what role can diet really play in supporting the immune system? In this episode of Living Well with MS, Geoff is joined by Overcoming MS Program Facilitator Karen Lee – retired intensive care nurse, nutritionist, author and recipe developer. Karen shares her MS journey and explains, in clear and accessible terms, how gut health, inflammation and diet are connected in autoimmune conditions such as MS. They explore dysbiosis and “leaky gut”, why fibre and the microbiome matter, and how whole-food plant-based eating fits within the Overcoming MS dietary recommendations. Karen also talks about her new book Healing from the Inside Out and shares practical, fatigue-friendly tips for eating more plants – without overwhelm. Watch this episode on YouTube. Keep reading for the topics, timestamps, and our guest's bio. 00:56 Welcome and introduction to Karen 01:49 Karen's MS journey: diagnosis, optic neuritis and early changes 04:46 From intensive care nursing to nutrition, writing and teaching 09:32 Understanding the Overcoming MS diet recommendations 10:36 Why diet matters for immune health: nutrients, fats, fibre and the microbiome 17:57 Dysbiosis explained – and how it relates to autoimmunity and MS 22:02 “Leaky gut”: what it means and why inflammation matters 23:16 Inside Karen's new book Healing from the Inside Out 26:46 How whole plant foods support overall health 29:50 Protein and plant-based diets: common concerns addressed 31:39 Practical tips for eating more plants and increasing variety 37:58 Favourite recipes, sauces and simple ways to add flavour 39:41 Batch cooking and freezing for low-energy days 41:26 Running the Taunton Half Marathon and fundraising for Overcoming MS Order Karen's latest book Healing from the Inside Out: Managing Autoimmunity with a Whole-Food Plant-Based Diet Support Karen's fundraiser for Overcoming MS Learn more about Karen's work New to Overcoming MS? Learn why lifestyle matters in MS - begin your journey at our 'Get started' page Connect with others following Overcoming MS on the Live Well Hub Visit the Overcoming MS website Follow us on social media: Facebook Instagram YouTube Pinterest Don't miss out: Subscribe to this podcast and never miss an episode. Listen to our archive of Living Well with MS here. Make sure you sign up to our newsletter to hear our latest tips and news about living a full and happy life with MS. Support us: If you enjoy this podcast and want to help us continue creating future podcasts, please leave a donation here. Feel free to share your comments and suggestions for future guests and episode topics by emailing podcast@overcomingms.org If you like Living Well with MS, please leave a 5-star review
Sexier Than A Squirrel: Dog Training That Gets Real Life Results
Send us a textWhat if changing what's on your plate could change how you feel, think, and even how you train your dog? We dive into a candid journey from a daunting fibroid diagnosis and surgery-first advice to a practical, food-first plan built around whole ingredients, simple prep, and flavour that sticks. Along the way, we talk about the 25% reduction that showed up on a scan, the meals that kept us going, and the mindset shifts that made healthy choices sustainable through packed training days.We get specific about what worked: ditching ultra‑processed foods in favour of vegetables, legumes, nuts, seeds, and natural fats; building quick wins like a 20‑minute lentil curry finished with lime; blending a blueberry‑coconut chia breakfast that sets up the morning; and keeping freezer-ready energy balls for the afternoon slump. Batch habits make the difference: slow‑cooker chilli loaded with greens, a soup maker that turns prepped veg bags into grab‑and‑go lunches, and simple hydration cues to separate thirst from hunger. For treats, we keep the joy without the crash—cauliflower nachos with guacamole, citrus‑coconut dessert bites, sweet potato fries, and a cashew‑based Caesar that tastes like the classic.Beyond recipes, we share why health upgrades translate to better dog training—more patience, cleaner timing, steadier energy, and clearer communication. Travel tips, sourcing strategies, and UK‑friendly healthy finds round out a plan that's realistic, affordable, and family‑proof, even for picky teens. If you've been on the fence about shifting your diet or wondered how to fuel long training days without relying on packets and powders, this is your blueprint.Ready to feel better and train better? Subscribe, share this with a friend who needs a nudge, and message us your first swap—what whole‑food habit are you starting this week?Support the showIf you're loving the podcast, you'll love our NEW Sexier than a Squirrel Dog Training Challenge even more! Get transformational dog training today for only £27!Want even more epic dog training fun and games and solutions to all your dog training struggles? Join us in the AbsoluteDogs Games Club!https://absolutedogs.me/gamesclub Want to take your learning to the next level? Jump into the games-based training membership for passionate dog owners and aspiring trainers that know they want more for themselves and their dog - Pro Dog Trainer Club! https://absolutedogs.me/prodogtrainerclub And while you're here, please leave a review for us and don't forget to hit share and post your biggest lightbulb moment! Remember, no matter what struggles you might be facing with your dog, there is always a game for that!
Can you steel an entire cantina? Not if it's owned by Cid! In this episode of Dark Side Divas we discuss the Star Wars - The Bad Batch episode "Infested" (s1e13). The Bad Batch return from a dangerous episode only to find out Cid is no where to be seen, and someone else is at her cantina. Why is Omega the voice of reason, again...in this episode? Listen to hear what the divas have to say.
Today's blockchain and crypto news Bitcoin is up slightly at $88,599 Ethereum is up slightly at $2,936 And Binance Coin is up slightly at $876 Bloomberg says Trump family fortune increased by $1.4B thanks to crypto Winklevoss Twins donate ZEC to support Zcash Hong Kong plans first batch of stablecoins Learn more about your ad choices. Visit megaphone.fm/adchoices
You know you need to create more video content, but between overwhelm, procrastination and overthinking, you're struggling. Let me show you how I create a month of content in hours so you can steal my methods and get out of your own way. This episode dives into how much content is required, batching strategies and how to keep innovative ideas flowing. –I'll create a profitable profile for you in minutes. Click to attract high-paying clients. https://go.taelerdehaes.com/bio-surveyJoin our Fit Pro Business Secrets Made Simple group over on Facebook for exclusive resources, trainings and help as you're growing your online fitness business. https://www.facebook.com/groups/fitprobusinesssecrets/ Follow Taeler on Instagram. https://www.instagram.com/taelerfit/Learn more about working with Taeler, whether you're just starting your online coaching business or scaling to multi-6/7-figures. https://taelerdehaes.com/
Do you avoid cleaning your house and then wonder when you last cleaned it? Do you struggle with an all-or-nothing mentality which prevents you from staying consistent?In this episode, we're talking about two ways to manage housework: batching tasks into big cleaning sessions and spreading small tasks into daily routines. Each approach has benefits and challenges, especially for ADHD brains and anyone who struggles with an all-or-nothing mindset.We'll explore why batching can help you use your energy and focus efficiently, but also why it can feel overwhelming or easy to skip. Daily routines, on the other hand, keep messes small and manageable, help habits stick, and require less motivation—but they can feel boring or easy to ignore if expectations are too high.This episode will help you understand how to use both batching and daily routines in a way that actually supports your home and your energy. You'll learn how to find a balance that reduces overwhelm, keeps your space manageable, and makes cleaning feel possible—even on chaotic days. If this episode blessed you, leave a review! Thank you so much! - XO COACHING Schedule a 15-Minute Consultation JOIN The Accountability Club FREE Daily Reset Checklist DO YOUR WILL @ Mama Bear Legal 20% Off with code: H&H20 MY FAVORITE PLANNER At-A-Glance Harmony Planner
After spending a decade working for the Empire, John Perkins walked away from his life as an Economic Hitman and gave up the game in his transformative book “Confessions of an Economic Hitman”. With the third edition of the book now available, we explore the role of the new group of financial arsonists who have set their sights on Latin America.Will China continue the process of empire building that the United States began half a century ago, or does its plan for a new Silk Road reward cooperation and collaboration instead? With the majority of its mineral wealth locked down inside the ground, could China secure the resources that it covets in South America while also raising the standard of living for an entire continent? Not if the American Empire has anything to say about it.—Guest Links John Perkins - Confessions of an Economic Hit Manhttps://johnperkins.org/—Watch the video version on one of the Macroaggressions Channels:Rumble: https://rumble.com/c/MacroaggressionsYouTube: https://www.youtube.com/@MacroaggressionsPodcast—MACRO & Charlie Robinson LinksHypocrazy Audiobook: https://amzn.to/4aogwmsThe Octopus of Global Control Audiobook: https://amzn.to/3xu0rMmWebsite: www.Macroaggressions.ioMerch Store: https://macroaggressions.dashery.com/Link Tree: https://linktr.ee/macroaggressionspodcast—Activist Post FamilyActivist Post: www.ActivistPost.comNatural Blaze: www.NaturalBlaze.com—Support Our SponsorsAnarchapulco: https://anarchapulco.com/ | Promo Code: MACROC60 Power: https://go.shopc60.com/PBGRT/KMKS9/ | Promo Code: MACROChemical Free Body: https://chemicalfreebody.com/macro/ | Promo Code: MACROWise Wolf Gold & Silver: https://macroaggressions.gold/ | (800) 426-1836LegalShield: www.DontGetPushedAround.comEMP Shield: www.EMPShield.com | Promo Code: MACROGround Luxe Grounding Mats: https://groundluxe.com/MACROChristian Yordanov's Health Program: www.LiveLongerFormula.com/macroAbove Phone: https://abovephone.com/macro/Van Man: https://vanman.shop/?ref=MACRO | Promo Code: MACROThe Dollar Vigilante: https://dollarvigilante.spiffy.co/a/O3wCWenlXN/4471Nesa's Hemp: www.NesasHemp.com | Promo Code: MACROAugason Farms: https://augasonfarms.com/MACRO—
Alex and Stephen start 2026 with a visit from Lord Maris for the seventh Barleywine is Life episode with four Barleywhales that have been lighting up the tradeboards on the secondary market. Featuring barleybobs from The Veil (STARVE: Exhibit H), Goose Island (King Henry II), Anchorage Brewing (Penta Oaked A Deal With the Devil), and Half Acre (Bazalt Wilderness of History). In the Beer News, Jim Beam temporarily shutters a distillery, Disneyland offers a $250 adult beverage, and Bells Brewing tests the spelling skills of their fanbase for this year's Hopslam release. To get involved with the "Life" International Barleywine Collab, click the link for info about the recipe, BSG discount, and links to help raise awareness of colon cancer. If you'd like to make a direct donation to help support Alex, head over to his GoFundMe. For more info about colon cancer and to help support the fight against it check out the Colon Cancer Foundation. Head to our Patreon for weekly exclusive content. Get the Malt Couture Officially Licensed T-shirt. Follow DontDrinkBeer on Instagram and Twitter
Hera is running from The Empire for the first time, with Chopper at her side! She's gonna have to get used to it. In this episode of Dark Side Divas we discuss the Star Wars - The Bad Batch episode "Rescue on Ryloth" (s1e12). Clone Force 99 gets a call for help from Hera, because Omega hooked her up with her personal cell number. Will Hunter be able to help Hera save her family from the clutches of The Empire? Listen to this episode to hear what Stef and Chris have to say. Warning: We do discuss a lot of real world politics in this episode. Stef and Chris have a lot to say, and if you are looking for an escape from the horrors of the world, you may want to skip this episode.
The semifinals of the College Football Playoffs are upon us. Mike and Jim pick both games. They also go through Wild Card Weekend in the NFL. The guys answer your mailbag questions and also wonder if the Giants are still a premier job in the NFL. All of this and more on the latest episode of Cash the Ticket today. To learn more about listener data and our privacy practices visit: https://www.audacyinc.com/privacy-policy Learn more about your ad choices. Visit https://podcastchoices.com/adchoices
© 2020-2026 Ivy Podcast Discovery LLC
