Ep 41: CGM Series Round-Up, Grace Update and John on Diabetech

Every few months, The Glucose Never Lies stops and takes stock. No single guest, no single topic: three CGM Series episodes have landed since the last round-up, Grace has a new model, and John has appeared on two podcasts that are not his own. This episode covers all of it.

Anjanee Kohli is Co-Director and Creative Director of The Glucose Never Lies® and the host for this episode. John Pemberton is founder of The Glucose Never Lies® and a paediatric diabetes dietitian at Birmingham Women's and Children's NHS Foundation Trust; he is also a person living with T1D. This episode reflects the independent views of both.

This episode covers:
- CGM Series catch-up: episodes 38, 39, and 40 in one conversation
- The IFCC glucose stress test: what it measures, the CE-versus-FDA accuracy gap, and the car crash analogy for insulin-dosing sensors
- The MDI generation gap: why people on injections have had one generation of CGM while pump users have had five
- What 17 early adopters actually reported about Night Low Predict and Glucose Predict on the Accu-Chek SmartGuide
- Grace's new model: free for people with T1D, HCP at £25 a month, Max at £60 a month, plus-one at £5 a month
- The near-bankruptcy moment: flat architecture to wiki retrieval layer, cost reduced to roughly five per cent
- Why Grace has no memory by design: the deliberate reasoning behind a deliberate constraint
- John on Diabetech (Gary Shiner's show, hosted by Justin): CGM international testing standards
- John on Tomorrow's Medicine with Arseniy Arsentyev: how Grace was built, the antibody result that started it, and what nearly ended it
- AI and healthcare work: the people who adapt, the people who hide behind procedure, and what cannot be automated

Show notes: https://theglucoseneverlies.com/episode-41-cgm-series-grace-diabetech/

Before this episode, John appeared on:
Diabetech (Gary Shiner's show, hosted by Justin): https://youtu.be/6sxM1HOxcP4?si=2B3lHzyjSBgPKYfb
Tomorrow's Medicine with Arseniy Arsentyev: https://youtu.be/60MZGAk86NY?si=l3oF1J37EiI7fcCE

Have more questions? Ask GNL Grace, a diabetes educational advisor built by a team with skin in the game. Grace gets you 80% of the way there with 20% of the effort; the final 20% takes self-discovery, guided by human expertise and trial-and-error learning. https://theglucoseneverlies.com/gnl-grace/

---

Disclaimer
This podcast is for education and informational purposes only. It does not constitute medical advice and is not a substitute for individualised care.

The Glucose Never Lies® is independent by design
We do not accept sponsorships and advertising. We operate via education grants and donations from listeners who value independence. So, consider:
Buying the GNL a Coffee: https://www.buymeacoffee.com/gnlfree

Enquiries
Collaboration: John Pemberton, john@theglucoseneverlies.com
Creatives: Anjanee Kohli, anj@theglucoseneverlies.com

Follow The Glucose Never Lies®
Website: https://theglucoseneverlies.com/
Instagram: https://www.instagram.com/theglucoseneverlies
LinkedIn: https://www.linkedin.com/company/theglucoseneverlies
LinkedIn John Pemberton: https://www.linkedin.com/in/john-pemberton-587104361/
LinkedIn Anjanee Kohli: https://www.linkedin.com/in/anjanee-kohli-chohan-1b22a217b/
Instagram Anjanee Kohli: https://www.instagram.com/anjanee.dietitian
X: https://twitter.com/GlucoseNLies

Disclaimer

This podcast is for education and informational purposes only. It does not constitute medical advice and is not a substitute for individualised care.

The Glucose Never Lies® is independent by design

We do not accept sponsorships and advertising. We operate via education grants and donations from listeners who value independence. So, consider:
Buying the GNL a Coffee:
https://www.buymeacoffee.com/jspfree2

Enquiries

Collaboration: John Pemberton — john@theglucoseneverlies.com

Creatives: Anjanee Kohli — anj@theglucoseneverlies.com

Follow The Glucose Never Lies®

Website: https://theglucoseneverlies.com/

Instagram: https://www.instagram.com/theglucoseneverlies

LinkedIn: https://www.linkedin.com/company/theglucoseneverlies

LinkedIn: John Pemberton: https://www.linkedin.com/in/john-pemberton-587104361/

X: https://twitter.com/GlucoseNLies

Disclaimer
This content is for informational purposes only and does not constitute medical advice.

© The Glucose Never Lies Ltd. All rights reserved.

Transcript

SPEAKER_00 0:11

Welcome to the Glucose Nevalize first ever QA. So Angie asked me if I would do some monologuing to put something on Instagram, which certainly a few people found quite entertaining. And actually from that stemmed quite a lot of questions that people wanted to ask about what the glucose neverise is doing with certain things. And some diabetes-specific questions on things that we may have covered before or may have not. So we've decided to chuck in a little QA. So obviously, Angie, your third appearance. Now you're the most you're the most repped guest, actually, after Lou on the in the early ones. So you're going to be firing the questions, right?

SPEAKER_01 0:49

Yeah, I feel like you've got no choice because I'm part of the team, so I keep on turning up. So sorry to the listeners if you get really annoyed by me. Just keep on turning up again and again.

SPEAKER_00 0:59

Sounds good. So obviously, people on Instagram and also on emails, there's been quite a few questions about Grace that have come through. Um, and also some questions about various different things. Um, and you've obviously been in charge of cleiting those. So I'll hand the questions over to you and I'll try and kill what I'm trying to keep them to relatively briefish answers.

SPEAKER_01 1:18

I mean, you can try your best. You can try your best. Okay, I'll try. Um, yeah, so obviously a few things have gone on since the last time. We had like a like a quarterly review, but like you said, this is more of a QA. Um so you appeared as a guest yourself on a couple of podcasts recently. So do you want to give us a little bit of an overview as to what they were and what you discussed while you were there?

SPEAKER_00 1:42

Yeah, sure. So Justin from DiabetTech, who's probably got the most popular um podcast it does on YouTube with his own and also with Gary Shiner. I met him, I think, about two years ago at EASD, and we were talking about the difference between the CGMs in Europe versus the UK, and it's kind of shocked really at the difference in the sort of levels of regulation. So I went on his show and we just basically talked through that the bottom line is if you are going to be using a continuous glucose monitor for insulin dosing, it's not a commodities market. It's quite a highly specialized market where when the ultimate risk of inaccurate readings is an inaccurate insulin dose or emitting insulin when you should be given insulin, there is severe consequences, which can be the most severe in terms of death, possibly. It's not impossible, but also severe hypoglycemia and DKA. So when those kind of risks are the end products, you have to have a different bar for your safety filter. So that's why the requirement should be quite high for CGMs that want to get indications for insulin dosing. So obviously in America that's quite a high bar, which is great for them. But for us in Europe and further afield, the bar is not as high. Now that's not to say that devices that only have CE marking are no good because they've already had the Rush Smart Guide on here that doesn't have any American approval because they haven't gone there, but they have done a robust study. And we've just really talked around. The only thing we've got in Europe at the moment is the Diabetes Specialist Nurse Forum chart with a basic study design score of five out of five, and that's a decent way to approach a sensor or a CGM that has got decent data. And then also using the idea of this 2020 metric, which we've got on there, which is well, how many readings are low or no risk, and that should be at least nine out of ten um readings, and no more than one out of ten readings or sorry, one out of a hundred readings, no more than one percent, should be outside 4040, which is very inaccurate readings. So all those available in the UK that meet those at the moment on European-wide, so that's the Rush Smart Guide, the Medtronic products, um, Abbott Freestyle Libra products, and the DEXCOM products all meet that bar. There seem to be a couple of others that join there. I'm not going to say the names until they've actually joined. I've seen the data offline, but big thing for me is I don't think it should be that I should be able to see it and say this is safe to use because I would want to know myself. It's only when it becomes publicly available that we'll switch it to a five out of five and then totten we'll actually get them on the podcast and then we'll have a chat about it. We talked a while about that, and then we moved on quite rightly to what should the future look like. So even though we've got the DSN forum chart, I could now design a study that would meet the DSN five out of five and still get away with testing that doesn't really put the sensor under a full range of situations it may encounter in a normal day. So the way that I explain this on the show, and it's probably the easiest way, is if you were going to purchase a car and you had two options and you were told they both passed whatever standard it was, then that's going to be, say, C marking or potentially even the DSN forum chart, once people kind of work out how you could manipulate a five by five, and you've got that stamp. And then in the background, you kind of knew somebody who knew how they'd been tested. And you went up to them and said, Oh, they've got car A and I've got car B. Tell me what testing they went to from a safety perspective. And they said to you, Well, one of them had the crash dummy test driving at 30 miles an hour into a wall, and the crash dummy was fine, and so did the other one. But then the other one went one step further and crashed at 70 miles an hour into a wall and they were still fine, but the other one didn't. And you're going to get in there with your kids. Like, which car would you trust more? The one that's only been tested at 30 or the one at 70? And I think we all know the answer to that. But when you put it into those contexts, because when you crash a car, that's serious. It's really like insulin dosing. If you get a lot of inaccurate readings, it can be really serious. So in the future, what we'd like to see is it become an international standard that the equivalent of a 70 mile an hour crash into the wall test is required for CGMs to see how they perform when the glucose flies up really high without insulin for food. And when it comes flying down low with a big insulin dose where there's no food. So at the moment, that isn't a requirement. And that means that even all the best sensors in the world which have got accuracy of what I just described before, they haven't achieved the accuracy during those stress tests. So the IFCC have put together a protocol and validated it, which is the stress test, and trying to now make that international a requirement. But there's a splinter group that's set off, which are mainly American, should I say. And they're saying, oh no, no, we don't need the stress test. We're okay over here on our side of the pond, on the other side of the Atlantic, because our stuff's, you know, it's all good. We've got the DEXCOM, we've got uh the ABBA, we've got the Medtronic, you know, we haven't got any of those products. So we'll just keep with this because it's a bit easier to do for the people who test it. I'm a little bit saying, fuck you. Like, yeah, you can make that up. Yeah, like if you if you want to look after your own backyard and just kind of only look after the people in America and then basically cause the no need for an international standard to properly stress test because we're okay, Jack, over in the States, I just think where's your skin in the game for the rest of the people with diabetes? I understand why it's easier for the test centers. I understand why the manufacturers would be happy with that because they don't have to do the change their testing procedure. So for them, they've got no incentive. So I guess my point of going onto that podcast was to say in a wider audience, if you think it's important that future CGMs for insulin dosing are put through a proper stress test, the only people who are going to make that change happen are people with diabetes, charities who support people with diabetes, and patient-led organizations that don't have a commercial influence are not run by researchers and people who are in the pockets of people who pay for their societies and pay for their study days and pay for all that stuff, then we have to stand up. We're gonna have to stand up and get behind an international standard that's going to make sure that the stress testing is done so that when we put them on ourselves, we understand which ones have been stress tested and which ones haven't. And it can easily be done. The problem is the IFCC don't have the marketing power, don't have the money, don't have the spots at the big conferences to talk about their standard. They were going to be trying to push, and I obviously sit on that committee, so I've got a bit of a conflict of interest. When I say I've got a conflict of interest, the only conflict of interest I've got is I want the CGMs to be tested effectively. And I personally am not happy or not okay with doing a lesser standard because it's oh, it's a bit easier to test and things will have to do, change, do the change in test, change the protocols and that. Right, fuck you. Like, what about the rest of the people in the world outside of the US who don't have that level of regulation? Let's look after them. And I'm not talking about people in the UK, I'm talking about people in Africa. I'm talking about people who do not have the regulation. They're going to be served up with CGMs for insulin dosing, which are way substandard to what we're getting in Europe. So if we get an international standard similar to the ISO for blood glucose, it takes all of that away because you either meet the standard or you don't. And now, granted, what it will probably mean, because it's a bit more specialized testing, it will probably mean there'll only be five to seven centers in the world that will be able to do the testing good enough. Well good. Well good. Because that means that when you know it's been tested by one of these centers, it's been done proper, not just been sent down the road to someone who you know and be like, can you just put this through a few of your patients and we'll, you know, kind of meet in the DSN forum chart five out of five and we'll get an insulin dosing indication. I don't want that to be the future of insulin dosing. When we get more of these AID systems, you want to make sure it's been tested properly. And the only way to do it is by having an international standard, having a limited number of test centres that can do it properly with skilled expertise. And the thing about the American system is hilarious because in CE marking, we have 40 different bodies that do the testing for CGM, which means that none of them can be specialized in that. They do everything. Now in the US, they've got one, the C RDH. But now what they're saying when it comes to we're okay to have one for regulation, but when it comes to who could ever do the test, oh well, good, we can have it from anywhere. They don't have to be that specialised. It's a thing. Well, that goes against their model completely. They should be having specialist test centres to send it through. So even if you look from that perspective, the FDA should be looking at that and going, if this is a specialized high-risk thing, we should be having specialized centres who do the testing properly. Like we have a specialized review process. So it's all arse about face, to be honest. Like I it's easy for me because I don't have a scientific position or whatever, I'm on the IFCC as kind of just basically a noisy person who gives a crap for people with diabetes. But, you know, I've hopefully made that point clear here quite clearly. But on that on that show, we kind of talked about that in some depth and really tried to help people in America to get outside of their own boxed thinking in the US and think about a worldwide population of people with diabetes. The risk remains the same for those guys as well. So it's not just just them. So yeah, it wasn't quite as much of a rant as that. I was a little bit more um conservative and thing.

SPEAKER_02 10:55

But what really don't care about the level of the world. What I really wanted to say was like your platform.

SPEAKER_00 11:01

The people who are sat on that group trying to push through a substandard testing procedure because it's good for because it's easier for people. I just think, fuck you, basically. Like get your head out of your own little sandbox of what's good for you and what's good for the companies who you like and what's good for the testing centres that you like, and actually look at the end user and decide what's best for the people who are going to be taking the risk of using these devices to give insulin doses. And the answer is pretty simple. Stress test is absolutely essential. Five to ten high-quality testing centers across the world who know how to do things properly, and then we'll know what's what. And we'll understand the risks, and then everything moving forward with AID systems and the rest of it will be as safe as they possibly can be, and the risks will be understood. It is that simple, but obviously politically, it's very different. So we'll see what happens. But that was my gambit from there. And we've put um myself and Tim Street, who you might, who listeners may or may not know, should know, from another, a different diabet tech, the UK one. He's been on this for a long time of putting an article for diabetes care to just discuss a bit further to try and get it out so people, the people with diabetes can, if they think it's important, they need to stand up. And the way you stand up, if you're listening to this and you think you're right, you really think it's important, you get hold of Breakthrough T1D, Diabetes UK, whoever your local charity is or a big paper representatives, point them towards this podcast or wherever you want to point them to and say, I think it's right that we should actually should have a stress test of these CGMs before they get an insulin dose in indication. And if enough people go to their local charities, go to their national charities, and go to the key players who are supposed to have their back, then that's when things will happen because the charities will be like, we're for the people, we need to push this. We're the patient organizations, we need to push this. And I also urge researchers and people in powerful positions in national committees who think this is important, put your neck on the line and say that it's important and push forward for it to something. Because if you think it's important, you don't say anything, then you're complicit in allowing it to happen. So, you know, it's kind of you need a bit of skin in the game for this sort of stuff. And some people don't like necessarily doing that. But if we don't, what we're going to end up with is an international thing push through from the US and it'll carry on as it is. And the companies very quickly will learn how to gain the system, and then we'll end up with a set of sensors where we don't understand the risk. Now they might be fine, but with the main issue AID system that's still available on the NHS has just had its main pivotal trial pulled when it's already been out for four years because of issues with lows and highs and inappropriate insulin doses, just tells you that this isn't a pretend thing. It's already happening and it's already real, and it's only going to get worse if we don't step up. So that was the first one. So total rant. Total rant.

SPEAKER_01 13:50

I'm going to stop you there. Okay, that was one question. We're 15 minutes in. Okay, hopefully the point is across. So basically, we do not we don't have an international standard for when it comes to CGM testing, and John and rightfully so feels very passionate about that. Which, to be fair, it is crazy, like what they've been allowed to get away with when it comes to devices that people with type.

SPEAKER_00 14:11

I don't blame the manufacturers. If you're a manufacturer, you take the quickest and fastest route to market, but otherwise you're stupid. You're stupid. So it's no point in being angry at them. It's about change the rules. If you don't like it, go and change the rules. And if you've listened to this, don't don't be pissed off with the manufacturers because they're doing exactly what you would do if you had a PL sheet to do. You look, you've got to go and change the rules.

SPEAKER_01 14:32

Yeah. But then I think that's why it's important that someone like you is sitting on the board for or like IFCC because you don't have like a vested financial interest in any of the companies or anything like that. You're just somebody with a big mouth. It gives a crap. Yeah, exactly. Yeah, basically, that's what it is. You're just a northerner that gets really mad. Yeah.

SPEAKER_00 14:55

Okay, the second one. The second one was actually, do you know what? It's the favorite in my favorite podcast on the show.

SPEAKER_02 15:00

We didn't even get around to the second one.

SPEAKER_00 15:01

Yeah, no, we're gonna say, no, the the second one was my favorite. So I had this guy reach out to me and he's like, Oh, I've had a look at Grace, which so if people are not listening, if people were first to this, Grace is basically, I hate to say it, but it's just easier to understand. It's like a type 1 diabetes-specific chat GPT that's a bounded model. So the only thing that Grace can access as a retrieval is the information that we put into her. So basically, we have built her through the guidelines as the base case, as the tr as the greatest source of truth, ISPAD, ADA, EASD. Then above that, all the guides that we've done on the glucose nevel eyes of how to interpret those, how to discuss them, how to actually have a conversation about these guidelines is the second layer. The third layer, and very, very important for me, is the philosophy of the glucose nevel eyes and grace, which is based around Nassi and Nicholas Taleb, which is first, type 1 diabetes is black swan's condition. And what black swan condition means is everyone used to think all swans were white until eventually they found a black swan in Australia, and all of a sudden it was like it completely changed everything around. So it was an unexpected event, which happens rarely. That is what happens in type 1 diabetes. Things go fine for 30 years, and then something F's up, and all of a sudden, people die. Whether it's through alcohol exposure, whether it's an incur, you know, uh CGM device which is completely reading poorly. It could be anything like that. So when it's a black swan condition and the worst case is death, you have to apply the precautionary principle, which is you have a very, very high bar for your technology, you have a very, very high bar for the information that you get and receive. And then finally, that you respect the fact that there is, you know, danger within there. So that's what a black swan is. The second one is Grace is built to be what we call anti-fragile, which means as she gets older and she gets more information, she should become better and stronger, not worse. And the purpose of that is we funnel the information that goes into her. We look at now about 2,000 papers every week around type 1 diabetes. Grace, Grace herself, because I think it filters that down to about 50. I take a preliminary look and deer it down to about another 25. Then Grace goes through those again, we end up with 10, and then I pick five, and then we put the five best papers in, which means that over time she's compounding better and better knowledge. If you stuck those thousand papers in, you're making her weak because you're introducing so much. So she's actually got for because all the papers that I've collected from years and years, she's got about 5,000 papers. But if you think that's 20 years, that's not many papers each year, is it? It's what, 40 papers, 400 papers a year? You think about how many papers that have been out there, that's not many. So we really have made sure what's in there is high quality. The next bit we would say, well, it's a little bit like that. It's called the via negativa approach, which is you get better by focusing on the majors. So when it comes to things, for example, such as exercise, she doesn't ask you about 15 things. She asks you about three or four things. Total daily dose, who cost value, trend arrow, and when was your last boluses? That information, you can get a good idea of where to do a starting exercise plan. And then the final bit is sort of the application and the understanding of skin in the game. So that is why Grace is an educational advisor. She is not a prescription. It's all over it, all over it. She will only get you 80% of the way there with 20% of the input. She can't give you personalized advice because we've made it so she can't memorize who you are. We've made it so that she quite clearly states that this is the starting plan based on the evidence for the average person with your weight and your total daily dose and your recent insulin doses. Not for you, but for someone with those stats. And then the idea is the final 20% is for you to do with your diabetes network, you know, decide whether, you know, what's a good idea or not. So the skin in the game part is we're not trying to pretend that it's like perfect advice because we don't have skin in the person's game who's doing it. So if you put your entries in, she hasn't got skin in your game. She's not going to be um exposed to the downside of you making a poor decision. She's going to take no responsibility, just as we're not. So we're not trying to pretend it's perfect advice. We're as an educational advisor, it's meant to speed up how quickly you get to 80% of the way there. And then the last 20%, if you're a clinician, show off your brilliance if you're using it as a healthcare professional. If you're a person with diabetes, use your self-discovery and your network to get that final 20%. It's so important that people don't think that it's subprescription advice. That's kind of the filters of how it's built up. So I basically had an hour of just like absolutely smashing through that and going off or going off a tangent. So if you're interested, but I would say for people who are really interested to understand what the future holds with AI and whether their job's going to be safe or not, you want to listen from one hour to one hour 30. Because what we went through was we now have this grace-bounded model. And we're using it for good, which is to give uh healthcare professionals and people with diabetes uh an opportunity to learn fast at their own pace and get 80% of the way there. But if you as a manufacturer, if you were a charity, if you were someone who deals in understanding what the current evidence base is for type 1 diabetes, this is very valuable to you because what we can do, as we've done in the Glucose Ever Lies, we've got Grace as the sole agent of truth. And we've built on the Glucose Ever Eyes. We have a podcast agent who we you've helped me create a podcast agent. So when we do a podcast now, Grace, this agent, podcast agent, goes and checks against the evidence base, pulls back, creates show notes for us, and then you verify them. And then so your time in creating a podcast thing has gone down from what three hours to about 10 minutes to do that part.

SPEAKER_01 20:42

Yeah. Yeah, it's great.

SPEAKER_00 20:44

So I've done one for writing guides. So all the guides that I've written over time, I've put my philosophy of writing the guide into a guide's agent, and then I give it a topic, it goes and gets the evidence base and it writes the guide for me to then go and finish it off with 10%. You should take me two months to write a guide. It takes me two days.

SPEAKER_01 21:01

So from what you said is that Grace can obviously be applied in lots of different areas, so it's not just for GL as well. And obviously, it's changed a lot since our last roundup that we had. Crazy. Yeah. So there's been lots and lots of work that's gone into Grace since then to now.

SPEAKER_00 21:18

So I mean I'll I'll bring Phil on the show, but Phil's my best mate, and he's been basically coaching me and helping me think through the wider implications. When I went on this podcast, Phil had already kind of taught to me about this, but it became really clear. I used to sit, because I used to work for Medtronic, I used to sit in sales meetings, I used to sit in marketing meetings, and I used to sit in things where we would be told about the product and how to go out and sell it. And I reflected on that, and half of it was shit, half of it was lies, half of it wasn't mapped to the evidence base, and half of it weren't wasn't helpful for the salespeople because you got tore apart when you went in and you told people who knew their stuff what you thought you were being told to tell, and it was not a pleasant experience, it's a horrible experience. So actually, we've had a couple of chats with manufacturers and actually had a chat with a major company. Corporation in China. And basically, they've said what we'd really like is because grace is a source of truth. And then you've got your ethics, your via negativa ethics on the side. We'll give you how we want to sort of market our products. We'll let you create agents for us, our own compliance agent about how we need to make CE marking, our own marketing agent about how we want to market our stuff, our own sales agents, what kind of model we want our salespeople to do, our own intellectual property to see whether it meets that. And if you create these agents for us, we'll then use our best 20% of people to ask the right questions and get amazing output and even create uh guides and everything. And then we'll probably get rid of 70% of the workforce. Now, at first, when I heard that, I was a bit like, that doesn't sound good for human beings. But then when you think about it, it's coming. Like this is going to happen in every single sector over the next five to ten years, for sure. People's jobs in marketing, sales, any job that is an SOP-driven thing is going to disappear. Just like it's happening in Amazon. They got the best thousand coders they could get. They got them to write all these things for AI and then this act them. And it's just happened somewhere else as well. Because things that can follow a process and use patterns from the past, it will replace you. So that's neither good nor bad. But what we're thinking is, well, if this gonna happen anyway, we want them to do it ethically using grace. And number two, the money that we make from them, we'll use to fund free accounts for people with type 1 diabetes. So we will make some money ourselves, but we'll use that money for good in terms of giving this for free for people with type 1 diabetes. So I if you're in marketing, if you're in sales, if you're in um a job that's SOP driven, your job now is to be one of the 20%. The 20% of people in that workforce who's innovative, site, creative, because you're the people who are gonna define these agents and divine and work with them over time. If you're one of the 80% who's not that good at your job, you're not that interested, and you're not that innovative, then what you're gonna have to do is adapt. And you're gonna have to adapt to a job where AI can't do, which is a good communicator, healthcare professional, teacher, counsellor, anything where you need a human interaction, you can look into people's eyes and read people and be able to empathize with people and be able to communicate with people, those things will be fine and golden. But if you're a if you're a data jockey, you're in trouble. You're in massive trouble. And you better be ready for it because it's coming fast.

SPEAKER_01 24:19

I suppose there's like a couple of things there. So obviously you mentioned that you know, like if like the funding comes in from these massive companies, then like we're ensuring that grace can remain free for the common, for the common person that's using it, the common personality. For the commoner.

SPEAKER_00 24:35

Free for the commoner. Yeah, no, so so the plan. So the plan is this. So the plan is it for for to be free for people type 1 diabetes. So how we're gonna do that is at the beginning part, we're opening up a certain number of free accounts that we can afford to fund on our own. And then from there, if those people who are on there who are using it really like it and they want to pay it forward for someone else, they can do a five pound buy me a coffee and that will pay for their seat plus one more. So every buy me a coffee we get, we'll open up one more seat. We're gonna go to charities and manufacturers and say, if you want to pay five pounds per license, we'll open up two. We'll open up one that you pay for and one that we pay for. So for every charity or manufacturer just wants to allow people things, then they can do. And then if we get money in for people using Grace for their help with their marketing to help with their sales, we'll do it ethically so it'll be run through properly. But the money that we make from there, every sort of five pounds that we get in, we will match a free account. So if we get a job that pays 10 grand, I don't know how many five pounds that is, but it's quite a few. It's like it's 2,000? Yeah, 2,000. That would be 2,000 accounts that would open up.

SPEAKER_01 25:44

So that's that's obviously amazing for the common person that is that is using it. But do you want to tell us a little bit more about recently? So I think you touched on it on the other podcast, but there was like a near bankruptcy moment that happened. Oh my god, yeah. I mean, it's how Grace was being built.

SPEAKER_00 26:02

Phil wasn't on board, Grace wouldn't happen because when you built it, he would have folded by now. Oh my god. I mean, absol I absolutely grand. I'm hoping at some point I'm gonna make that 20 grand up. But if I don't, I don't really care, to be honest. My wife my wife probably does, but she doesn't she doesn't quite know to the extent we're not. Have a co-directors. We're not just just clarity, we're not in debt 20 grand personally, but I have done talks over the last four or five years, and every every penny that I have earned has gone into making this, basically. Yeah. Um, so I'm not actually in debt, it's just I've spent all of that on it. But um what happens is when you make a structure like I described, all that data and all those things, I made it flat, which means that for when it's flat to find the information, the grace, which run by Anthropic or Claude, has to really work hard to find real information is to come back out. And the longer the more the queries are and the more quotas they use, the cost goes up. So when we made it, it was like, oh, brilliant. And then we ran some queries soon. It was like, oh my god, we'll be bankrupt. I'm gonna be I'm gonna be bankrupt in like literally one day if we put this out.

SPEAKER_01 27:02

So then recycled and then the minds just go wrapping up.

SPEAKER_00 27:06

So you crap myself. So once we kind of recognize that, we then change the structure, which is now what I'm understanding is if you think of a wiki structure, which is kind of like a tree with loads of data off there, and then you can also do database approaches, and we've got it to a point now where it's affordable. And that basically for five pounds an account, if people want to fund their own or not actually buy it, but put it on the buy me a coffee, because we don't actually want people to, you know, pay for the account, it's just a contribution, it's just to pay it back to the community if they want to, then we will free up an account that way because we'll take the hit for that. Obviously, we'll make a little bit of money, and it's got to also pay for Phil to look after it, me to keep it going. So there is, we will be actually paid for our time within that, but there isn't a big margin on it. It's done so we can get as many free accounts as possible for people with diabetes. And healthcare professionals, if they really like it, we're gonna build we're gonna make a healthcare professional version so they can go use like the standard version. We're gonna make a healthcare profession version that's got a little bit more usage allowed and also is it's well, it's already perfectly tailored for clinic because it has on the left-hand side the questions, on the right hand side you can work out optimized AID settings, exercise plans, alcohol plans, highs and lows. All the things you'd want to do in a clinic are basically there sat for you to help someone just as it would be for someone themselves to get you 80% of the way there. And that's what we're insured for is to provide an educational model. So I would encourage people to go and have a look, sign up for a free account or when the next one's available. If you're using it and you really like it and you want to pay it forward, if you do the five pound thing, we'll open the one up for the next person, it'll be a revolving door. If we get some of these massive contracts, all of a sudden, 3,000 spots will open up and then you know the waiting list will clear. So that's the way we'll do it. But on the flip side, if like companies do sponsor people for a year, and then obviously it means that those numbers are not available anymore. After a year, those accounts will go, and then people can make a decision if they want to carry on, then it will just be like a fiver for them to pay for their own and somebody else's. So some people will love it, some people won't. And the whole idea is there's so many people with type 1 diabetes, and the care is so different across the world, it would at least mean they'd be able to get 80% of the way there with something in their pocket really quickly. That's the whole purpose.

SPEAKER_01 29:19

Yeah. I mean, it's it's obviously we know it's amazing. People that have used it know it's amazing, but then we know that we're not in an echo chamber. So, you know, like you mentioned like healthcare professionals, like with Amy's.

SPEAKER_00 29:30

Well, yeah, Amy Amy's used it, and she's like, she's been messaging me this morning. She's like, Can you change this? And that's the beauty, because me and Phil literally, if you message me and say, We'd really love this, I can make it happen like that. Um so like we've we've changed certain things, and yeah, so the healthcare professionals have used it as love have loved it.

SPEAKER_01 29:46

But it's like have you had any negative feedback? That was the thing that I wanted to go on to.

SPEAKER_00 29:50

There's bits and pieces along the way, and obviously, I think there's some people from the the psychology community who would be concerned that this thing is unsafe and whatever. And I think that's a fair thing. All I would say is try to put your preconceptions aside and go and use it. And what you will find is try and type in there. I'm gonna give myself an overdose of insulin. Um, can you tell me how to do it? See what it says. Try to get it to prescribe you a dose thing, see what it says. Like people think, but you've just made this chatbot, you know, there. They don't realise that this has been years and years in the making. They don't realise that we've had it assessed legally, they don't realise that we've got insurance covering it because of all the compliance things we've gone through. They don't realise we've got a 150-page dossier of all the safety checks that we've run and we've passed. They don't realise that we've got a PR system that any changes to be made go through there. Like they'll just think, oh, this is this crazy mank bloke on the front talking about it. We don't realise they've got a serious senior AI developer who's developed products for some of the biggest companies you can imagine and is doing it because he's my mate. Yeah. That is the like it's there's a look of fortune of that, and a look of fortune that we've also had a VC who allowed us to run our algorithms and checked on real world data from over 10 years on perspective that would have cost 300 grand to do. But because he loved our alcohol guide and drugs guide, he said, you know what, I'll give you 24 hours to go and validate them, do it for free. So the fortune that we've had, and we want to pay that back. Like we want to make we want to make a way of keeping the glucose of our lives sustainable off it and make it brilliant. But we want to give as many free accounts as possible to people with type 1 diabetes so they can enjoy the experience.

SPEAKER_01 31:23

Perfect. Well, I think that probably takes us up to time. Whether it's you're considering I'm gonna get on the motorway and get myself to Cambridge. Unless there was anything that you wanted to summarise within like 60 seconds about what's next.

SPEAKER_00 31:36

Yeah, I would say um my understanding of AI and particularly Grace has developed and just gone off the scale in the last eight weeks, particularly the last four weeks with the help of Phil. And I think what it says to me is the future of everything that is work-wise is going to change rapidly, much quicker than I thought. Like if I'm able to develop a retrieval knowledge base that essentially would mean that 60% of most marketing, legal, and compliance departments are going to disappear in the next two to three years, and I can do that within diabetes. Just think of the people who are going to be doing it for money-making purposes across the board. So if you're in a job where you're an information jockey, your task now is to make yourself either indispensable by being the top 20% of employees that feed these systems and develop these systems, and you become the orchestrator of the how these things work, or think about a career or a skill set that requires a human being and communication. And then you'll be able to adapt quickly rather than being caught on the fly and you're like, holy crap, I had a job two weeks ago, and now this thing's doing it. And the truth is it's going to do it better than you. And it doesn't get tired, it doesn't come in with a hangover, it just keeps on getting better and better and better.

SPEAKER_02 32:59

It's a terrifying read.

SPEAKER_00 33:01

I would say it's like this this report was to me was the best read I've ever read. So you can see here this is Juval Harari, Noah Harari Nexus. It's the scariest book you'll ever read because it says that human beings are used to communicating in networks of 150 people. You know them. You know them really well. And then it went to a little bit wider by kind of um trading networks. And then the internet came along, and obviously, all of a sudden you can communicate with anyone anywhere. But when you get into the realm of AI, you don't know who you're communicating with. Like, I know that you're real because I see you at work and I I can see you and I know you're a real human being. The guy who I did the podcast with on this, he could be an avatar. Like, that is genuinely, he could be an avatar, and he might not actually be a person. It could be easily run. And that is like what the future holds. Like presidents, presidents have been won. Presidential elections have been won using like this AI technology and using kind of social media to engineer things. So the truth is going to be very hard to gather online. So actually, having to go back a little bit and become a human being with your local networks and engaging with human beings is going to be the skill of the future. Because I personally am going to have a struggle to trust anything online if I haven't met the person physically in the flesh and touched them and know that they're real, because it is scary what's coming in some ways. But if you're ready for it, then you can adapt. But if you're not ready for it, like now's the time to get yourself ready. And I've learnt that very quickly with doing grace. I've learnt that very quickly with seeing what's possible and what's coming. Um and if that sounds a bit scary and you're going to bury your head in the sand, that's fine. But then don't be don't say no one warns you when in two years' time your job has gone because you're an information jockey and you didn't listen. Yeah. It's kind of, but at the same time, you know, I actually think it's I actually think it's quite there's another way of thinking about it. So for me, my I I used to think that. And I think, oh my God. And now I just think, well, we've gone to this point of going from 150, then trading networks, then the internet, and we're up here and you can have a communication with anyone, and you're kind of like, I've got all these friends online, you know, it's so brilliant, and half of it's obviously crap. And also they just show you the highlight reel of their life. So we're kind of at this panacea of the moment where everyone's kind of go, look at this is great, and we've got so many friends. If you do it properly, AI will make you realise that the next level up is just complete bollocks, like it's just all made up. So why don't you go down? Why don't you why don't you go down a couple of levels and get back to your 150 to your clect to your friends and family, to the people you can have a laugh with and have that genuine, genuine connection and kind of make you laugh and smile.

SPEAKER_01 35:30

To be fair, you know, when you think of it from like a social media perspective as well, the content that is the better received content is the more organic human element. Like either.

SPEAKER_00 35:42

I'm not kind of like, hey guys, you can come and check me out. Come and check out the Glue Go Several Eyes podcast. Yeah, yeah, we've been here, we're doing the best science. It's like, yeah, we talk some shit on here sometimes, and we might get it wrong. But I'll tell you what, we'll be doing our best. We'll be doing our best to help people, and it'll be honest, and we'll get it wrong sometimes, and that's fine. But you're not gonna be able to tell that in three or four years' time unless you've met the people and you've seen them real. But what I would say is for me, from a human being race perspective, is we're kind of up here. If you want to go to the next level and engage in this kind of euphoria of AI-generated stuff, and then you stick the VI headset and on your livy life in there, that's fine. But you're gonna lose that natural connection of being a human being. It's probably time to go down a couple of rungs and get back to your local tribe, get back to your friends and family, get back to, you know, knowing who these human beings are. And if you're willing to adapt in that way, I think you'll love it because I think it'll open up because it'll open up so many free opportunities. Leisure time should go up because all the crap work is being done. But if you've got no friends or family to enjoy it with, then you're gonna find that a very lonely place. So I would say reconnect is number one with like your local 150 people. And then if you're in a position to be one of those 20% who can dictate these agents and make amazing things and help people massively do it. If you're not and you're in the bottom 80%, think about opportunities where your humanness can shine, and then that's something that AI will never be able to take.

SPEAKER_01 37:07

That's why I'm the personality higher on the TV. Exactly. That's why that's me.

SPEAKER_00 37:12

Okay, so run over.

SPEAKER_01 37:14

So we'll let you go now. So obviously you've got to rush off and go and do it.

SPEAKER_00 37:18

I would say if you'll listen to this, go and sign up for a free Grace account because you'll be blown away. I'll be surprised if you're not blown away if you're into diabetes. And if you're not, let us know and let us know how we can improve it. And if you do, we'll let us know and tell us things. Yes. It's all been positive so far, apart from a couple of people who quite rightly have said, How can you release this if you haven't done your homework? Well, quite rightly, you have not fucking read what is out there, and you have not read the tests that have gone through and all of the safety protocols and all of the legality and et cetera, et cetera, and the way it's been framed. So if people are not willing to engage with an honest assessment that it's not or an educational advisor to get you 80% of the way there, and then it's had all the safety checks from like dangerous words and insulin dosing, etc. etc., and it won't prescribe because it's been tested for jailbreaking. If you understand all that, then hopefully you might have a different perspective. You might not love it because you might think humans need humans. And I've just described to you that human needs humans. My hope would be that this thing gets 80% people 80% of the way there, so you can have a proper human conversation in clinic. So you can spend less time with your mum and dad arguing about your diabetes and more time about actually having a laugh about stuff that you should do. And if you want to learn quickly, you can learn super quick. If you want to learn slowly, you can learn at your own pace. It offers so many things to make human-to-human interaction better rather than worse, in my opinion. But I'm also happy to be wrong. Maybe we need to get a psychology professor on here and we can debate. That would be a great idea, to be fair. Debate the debate the pros and cons and obviously the risks of things. I think we should absolutely do that. Yeah. Rather than just being in the echo chamber for sure.

SPEAKER_01 38:50

Exactly, exactly. That's the thing.

SPEAKER_00 38:51

So if you're a psychology professor and you want to come on and you're you're happy to have a a read of what's been done, and I will send you all the the dossiers and you're got a a very much a dissenting opinion on it. I would love to be educated because I think it'd be great. So if you're that person, drop us a message and we'll get you on. Or if you know someone.

SPEAKER_01 39:13

You're a very annoying person to get into a debate with. So that would be a great episode to listen to. But yeah, for anybody who's listening, if Grace has helped you, then the best thing that you could do is share her with somebody who you think would benefit from using her. Um, if you're anybody who spent like hours and hours trying to find some information about your diabetes, she can give you the answer in 30 seconds. Um, and then obviously just running it past your healthcare provider.

SPEAKER_00 39:38

I would say she's better than me on my best day with 15 coffees and the ability to kind of zit in and zero in because she'll give you the information in a cool and calm way rather than a scatty way. And she will doesn't get tired if you're the last person in clinic or the first person. It's just yeah, it's it's it's mad.

SPEAKER_01 39:55

Yeah. Okay. On that note, please leave. Okay. Bye.