Episode 38: Q1 2026 Quarterly Review, GNL Grace, the Explorers, and What Changed

Three months ago, GNL was a website full of excellent content that was genuinely hard to find. In Episode 38, John Pemberton and Creative Director Anjanee Kohli talk through what happened next: a complete website rebuild, five interactive explorers, the launch of GNL Grace, real-world validation on 1.5 million patient days, and what the Robin Hood model of type 1 diabetes education actually means.

In this episode:

• CGM guide update and the five-out-of-five accuracy chart
• How the five GNL Explorers work and how clinicians are using them in practice
• GNL Grace: a bounded AI educational advisor built on 2,000 curated citations
• Three Grace tiers: Basic (free), Pro (30 queries/month for subscribers), Max (manuscript-grade output)
• The Robin Hood model: industry grants fund free education for everyone
• Real-world validation: 33 assessments on 1.5 million patient days
• Why the skill of the future in healthcare is compassion, not knowledge
• Phil Hayes joining as Technical Director
• GNL merch and the Diabetes UK giveaway

Chapters:

• 00:00 Introduction
• 01:55 GNL Grace announcement
• 02:32 CGM guide and accuracy chart
• 05:30 Phil Hayes and the explorer infrastructure
• 09:40 GNL Grace: the six-layer bounded AI
• 10:50 Grace tiers and the Robin Hood model
• 13:31 Grace Max: manuscript-grade output
• 17:31 Real-world validation on 1.5 million patient days
• 19:58 The skill of the future is compassion
• 30:09 What went wrong and lessons learned
• 35:06 GNL merch
• 36:10 Closing

Links:

• Show notes: theglucoseneverlies.com/episode-38-quarterly-review
• GNL Grace: theglucoseneverlies.com/gnl-grace
• GNL Explorers: theglucoseneverlies.com/gnl-explorers
• CGM Series hub: theglucoseneverlies.com/10-13-cgm-series

Host: John Pemberton. Director of Creativity: Anjanee Kohli.

This content is for educational exploration only. It is not medical

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_01 0:11

Welcome to the Glucose Never Lies Podcast. This is the second of our quarterly reviews, and you'll see I'm joined by the creative director, Angini Cowley. How are you doing, Anj?

SPEAKER_00 0:22

I'm not too bad, thanks. How are you, John?

SPEAKER_01 0:24

Yeah, I mean we've had we've just had a disastrous start to this episode. I've spent about 15 minutes trying to get my camera on. We've finally got my earphone sorted, and then you couldn't hear me, so it's been all over. And then in like 40 minutes, I'm gonna get told off because I'm gonna run over going out for my mother-in-law's birthday. So yeah, autopsy.

SPEAKER_00 0:40

I told you, I told you we could rearrange, but that's on you.

SPEAKER_01 0:43

Yeah, okay, I know. I'll just take the hit. But yeah, we'll we'll we won't do it really quick, but we have we've got some interesting things to talk about. And things have really sort of taken off in a massive way in the last few months, as people will have seen with the website. So we thought it's probably a good time to just update kind of what's been happening, what's going on, why am I wearing a hat? Why have we got Dexter the dog in the background? Why have we got hats? So many questions. Yeah, there's quite a few things. So I'll hand over to you. Um, why are my teeth whiter than they were last time? I mean, there's all sorts of questions.

SPEAKER_00 1:17

You've got a way to go.

SPEAKER_01 1:19

That's going through the black hat. So go on. You you kick off, you lead us through this because uh I've got too many tabs open in my head. I just need to do some straight quick QA's.

SPEAKER_00 1:28

Okay, cool. So the last time we did an episode, me and you, was in December time. And then, like you said, quite a lot of things have happened since then. And we had some big plans for like the year that we discussed in that episode as well. So since then, we'd been to a TTD in March, which is one of the biggest technology conferences for diabetes. Obviously, the website has had about six facelifts, probably in the last two weeks. More than Sharon Asborne. Yeah. G and L Grace was born. So where do where do you want to start, really?

SPEAKER_01 2:04

I think the best place to start is what the plans were and how they've probably evolved slightly but also taken off on a rocket ship. So the things that was going to happen is the CGM guide update, which is basically done now. So the CGM guide has been updated. The key things being we have done a couple of episodes, one with Othma and one with the Diabetes Specialist Nurse Forum on the importance of getting study design right first, and that what most people will know is that the chart that we developed is a five out of five, and it still remains that only the key players are on there with the introduction of Roche, of course, with the Smart Guide. So you've got the pretty much the Dexcrum G7, the Freestyle Libre 3, the Roche Smart Guide, and soon to be two players coming. Um, ICANN and the CareSent. I've seen the data for those, and I am very confident that they will hit a five out of five. However, seeing is believing, and seeing as in publicly available data is the only time where we'll change it to a five out of five. So that who knows how long that will be. And but for the time being, it remains that those the key players that remain on there, and obviously the Medtronics in player as well. So we've had um the podcast with Adam from Dexcon, which is a really good episode actually. It really talked about the things that make them different and um also some honesty about some of the challenges they've had manufacturing-wise, which I think was great, because as I said on the podcast, when you're producing that many sensors, you're gonna have some issues. And if there's people cutting corners slightly, they got caught. And that's fine because one good thing about the FDA is they audit that thing, and you find out there's no there's no cutting corners. So that's one thing about having a FDA-approved CGM is you know you've got that as an accountability piece, and that's been shown with them. It's also been shown with Libra. And obviously, we did the one with the Freestyle Libra, which is meant to be going out tomorrow, but has just been held back slightly, and they're just dotting the I's and crosses and T's at their end. So we thought we'd hop in with this quick quarterly as a perfect timing to update. But again, that was a great session with Mike. He was brilliant actually. He walked us through what the product has to offer, but also what the company will offer the different cis the different healthcare providers. So that was really, really good, a really good session. I think people get a lot out of that. And we've got one coming up with Roche and Medtronic, and then later on, once the other two get their five out of fives, we'll get them on as well, which is fair. But what we're really excited about is as part of the facelift for the CGM guide, we've now got a CGM selector, which basically asks you out of eight things, what are your three most important things for a CGM? You click a three, and we've got a little cool algorithm in the background, and it spits out and it says these two will probably be the best for you given what you want. And then you can go and look at them. So instead of having to decide, like, and this is a good thing about hopefully, I think what the glucose never lies, like we major in the majors, what's the most important three things to you? And then let's just give you a CGM couple of options that work for that. You don't want to know L the whys and wherefores of everything. You just want to know, is being able to optionally calibrate massively important for me? Yes or no? Is the length of where a massive thing, yes or no? Is the size of it really important? Is the predictive alarms really important? Just pick the three that's most important to you and it will tell you which are the two best options, and then you can go and kind of pick from there. So hopefully that'll make the CGM piece a little bit easier. We only include ones with a five out of five, and therefore we help you select. And that hints to what's been going on in the background with the quality of the website now, in terms of the AI, the algorithms, the shape of it, the look of it has just absolutely gone bananas. And that's because my best mate has come on board as um clinical, sorry, it's technical director, Phil Hayes. And so Phil is my best mate, beautiful best mate at my wedding. I was the best mate at his wedding, and it's just fortune that with my guest clinical experience and education and publications, bits and pieces, we had, I think we counted 150 pages on the website, which is like a full, a pack full of content for but most people that's just ridiculously overwhelming. And I I kind of get that, but I don't know how to make that different. And Phil came along, he was like, Okay, so really what you want to do is when people turn up to your site, they want to ask very simply, they want an answer to something. They're coming to your site for an answer to something, but they're finding it's hard to find it at the moment, which is a bit of a bitter pill to swallow, but it's probably well, it's not probably true, it is true. And he was like, Okay, so I can develop something for you. If you tell me what the key questions are for exercise, if you tell me what the key questions are for alcohol management, if you tell me what the key questions are for an AID system to get the most out of it, you tell me what the tree is, and I'll build you some explorers where essentially you pop in three or four details that are personal to you, and it will give you what the evidence-based answer is for the person who is like you. So not the person who is you, because these algorithms and things cannot understand personal context. They don't understand your social circumstances, your particular sensitive to insulin, insulin sensitivity. But what they can do is say what the evidence would say for the average person, for example, for me, who weighs 100 kilos, who's on 40 units of total of insulin a day, and has got these preferences, this is the settings that would be neutral, pushing towards aggressive or strongest or weakest. For the AID algorithms, for example. So you can go on there and depending on which system that you're on, you can pop in those numbers and it will give you a scale of one to five of a set of settings that the average person would want for how aggressive or how strong the algorithms to be, and that's for all of them. That took a lot of work to put that algorithm together, but now it is so good. Like we've been using it in clinic. I've had loads of pictures from my mates who've been using it in clinic, and they've been like, which all we just do is pop in a few things, and it tells us for this particular system what are the main levers to tweak, how to tweak them. And for the average person, this is what they want, and this is the important thing. Some people would look at that and go, Oh, you're taking our job away. It's like, hell no. The important thing is we're getting you 80% of the way there, and the other 20% is where you come in. That's where your relationship with the individual comes in, your clinical skill, knowing that for this person they prefer this or they prefer that. That should be the future of good clinic consultations, not number crunching and finding out these bits and bobs. It should be, this is we're 80% of the way there, let's just do the polishing for the 20 and have a really good relationship. So I'll be eternally grateful for Phil for doing that. And obviously that then kicks on to the newest thing. So the explorers have been out for about two or three weeks, and people have been really enjoying those. But then Phil said, like, we're not stopping there, we're like, we're gonna do it properly. I was like, What do you mean do it properly? He said, Well, that's still quite a lot of effort to find the explorers. You just want to be able to rock up on the site, ask the question that you want, get a really, really good answer, and then point it to the best two places. Like, well, how are we gonna do that? He said, Well, we're gonna develop an educational advisor. He's like, What do you mean? He's like, Well, we're not gonna do a chat bot because a chat bot is just a whole load of decision trees. We need something that's built on a structure where you put a base layer in of clinical knowledge at the bottom, for example, the ISPAD guidance and the ADA guidance, and then you put all the stuff that you've built over 10 years, all the glucose levelized guides on how to frame clinical conversations. Then you put the deterministic stuff in there to sort of frame questions for that we've developed with the explorers, and then on top of that, you put good quality evidence on top of that for the different areas, and then you keep on layering on only the best evidence. So we've basically developed a six or seven layer educational advisor that you can go on, and there's three levels. So there's the standard Grace, so we've called it Grace after my daughter because she's pretty clever, she wants to know, she's very curious, and she's gonna grow and get stronger with time. So that's kind of why we called it GNL Grace. So it kind of got three levels. We've got the Grace, so anyone who turns up to the website doesn't have to sign up or subscribe, and they can ask three simple questions and get three really good answers and a direction to where to go. Now, if you're a subscriber, which most of the people will be listening to this, you're gonna get Grace Pro. So, Grace Pro, once you sign in, you will get a web page and it looks like a proper chatbot, like a Chat GPT or Amazon, Claude even, is my new best mate, and you can put in there really like pretty sophisticated questions, and it will give you three or four paragraphs of amazing content, references to go and look at, and the evidence grade behind the it behind the answers. It is freakishly unbelievable how good it is. I'm pretty difficult to impress in there, but Phil was like I can I will impress you, and I was like, Yeah, I'm not sure you will. But I looked at that and I was like, that is it's ridiculous. It's ridiculous how good it is. And so we've done it, and this is the important thing for me. I want to call this the Robin Hood model of type 1 diabetes education. That is 30 really good queries free for everybody who's site who's a member of the glucose novel eyes, a subscriber, every month. 30 queries of whatever you want. For most people, that will be more than enough to give them an amazing answer. Um, and they will be able to get answers to questions that's on the latest evidence, delivered in a way that it would be me delivering it because I've made it uh and in a balanced way. It will never tell you what to do, it will always be education only to get you 80% of the way there. But I mean, I've been playing around with it. It is, it's it's crazy. I honestly can't believe how good it is, basically. What do you think? I mean, you've had a bit of a play with it, what what you're doing?

SPEAKER_00 11:27

Yeah, yeah, it's amazing. It is really, really good. Like you said, it obviously it's it just gives people that have got type 1 diabetes because they don't have they don't have access to healthcare professionals all the time and everything like that. So it kind of puts the information in their hands, which is which is really good. But I suppose to summarize what you've said so far, so you said since the beginning, you've been talking, you've been monologuing so far. So we're gonna so from the beginning of the year, obviously we went to ATTD, the website's been completely redesigned from the ground up with the help of Phil and Claude. Um, and then GL explorers were first. So we've got some videos on social media that are coming out that are basically explaining what the GNL explorers do. And then the most exciting bit at the moment is GL Grace. Anyone, like you said, who is a subscriber or a subscribe to GNL has got access to her. We've personified her, so yeah, has got access to her. So yeah, that that's really cool.

SPEAKER_01 12:34

So this is something that you don't know because this has happened over the last like 12 hours. Obviously, Grace is like on the bottom right hand corner is a little button you press and you can just ask simple questions. But actually, when you log in as of Wednesday, when you log in with um with your username and password, there will be a dedicated page you click on. It'll be a full page. On the half page will be your questions that you answer, and on the other half will be the explorer. So you can dip in and out of both. And on the left hand side, there will be common commands that you will you can click in. So if you just want a summary on what's the ISPAD guidance on this, or what's the current state of evidence on GLP ones, or what's the current evidence on alcohol? Give us, you know, give us some if for a person like this, what would the plan be? And it would say for a person like that, the average plan would this, but then check with your healthcare professional. So we've been through all the legal stuff and made sure we've covered from an insurance perspective because it's not medical advice, it's just bringing the education to life. So that is amazing. But then Phil was like, we still need to go one step further. And I was like, What do you mean one step further? He's like, We need a great king.

SPEAKER_00 13:31

You've got a wife and kids, you've got a lot.

SPEAKER_01 13:34

Still, just at the moment. At the moment.

SPEAKER_00 13:36

Exactly. Exactly. Does Phil have a wife and kids? He needs to think about them too.

SPEAKER_01 13:40

But this is this is why it's so good at working with people who have seen other things in other areas because you need a Grace Max. I was like, what do you mean a Grace Max?

SPEAKER_00 13:48

You mentioned this. You mentioned this yesterday. Yeah.

SPEAKER_01 13:50

Something with basically the the superpowers that you would envisage for an educator, but also can create you unbelievable PDF resources based off what you want. Can give you the outline, and even if you wanted to, write you a publication level manuscript on an area. Can develop unbelievable HTMLs about click button ones that are really fancy, really cool, that have got all the evidence base within but delivered in such a cool way, cool graphics. If you've got a manuscript, you can drop it in, and if you want really cool graphics, it will make you the best graphics on your script because it's got a citation database of my best 2000 citations of diabetes education and management and extra expert, and it pulls on all of those in a layered way. I've been on and I put in a manuscript that would have been accepted and is a really good publication. I was like, can you just give me a peer review of this? And would you have done better with the graphics? And what came out, I was just like, that is unbelievable. It would have made the publication so much better because the graphics are just so good. Um so we kind of we developed a GNL Max, which is basically a Grace Max, which is for people who really want to go bananas and really do amazing things. For example, at work, we can redesign our whole diabetes education materials. They're already good, but to a level which is just kind of like will blow people away, was absolutely blow people away. So we're offering that to people. That obviously to the problem is when the sophistication goes up, the money that we've got from education grants from the companies will cover everybody for a free three goes on their website. If you sign up, 30 a month queries for theirs. So that's covered by when I say the Robin Hood model, us taking money from the companies to provide education for free, which is what I've been working really hard to get the money for. But when you go to the max level, the number of the word is tokens, the number of tokens it burns to create that kind of material is like fairly expensive. So we've had to price that for people with diabetes or anything else at£50 a month because the quality that it produces is just like so. I paid a chat GPT subscription for£22 and it's decent, but the graphics it turned out were turd. Like they were they were really not that great when you were trying to get them specific to diabetes, they didn't really have a clue. This thing is it's ridiculous, it's just next level, it is unbelievable. So I would say to people like most people will be absolutely fine with a pro. They will just you only answer questions. But if you're a healthcare professional, if you're a person with diabetes, if you've got a person, you're like a blogger and you've got unbelievable, you've got a lot of big following, you could make the best stuff you could ever imagine. And 50 quid a month is absolutely nothing to be able to do that, and you get like 35, 40 of the best queries of the like strongest things, and when you go on there, you can like there's tabs like manuscript, create graphics. So we've put down all the common commands that you'd really want to do. So we've already funnelled it into what people would want to do, and I've had a play with it, it's just ridiculous. So that is the bit that I'm most excited about. So that's gonna be out on Wednesday. So people are gonna hear this on Monday, and by Wednesday, we're gonna drop that, and me and Phil are gonna do a podcast on Monday and put it out on Tuesday so that people really understand what it's about and they can find out a little bit more.

SPEAKER_00 17:06

So we're gonna edit that podcast or a month.

SPEAKER_01 17:09

I'm I'm doing that one.

SPEAKER_00 17:10

I was gonna say that's news to me. That's an exclusive.

SPEAKER_01 17:13

Yeah, so it's gonna be well, there's probably spell if there's spelling mistakes and things in the um in the things you can it's not Anthony, but we need to we need to get that out before it launches because um people just understand what goes behind it and it's ridiculous. So I'll kind of finish on that bit on the GNL Grace, but for me, it's taken over the last three months, it has been a lot of work. I mean, it has been late nights, early mornings type of thing, but what is on there now is just ridiculous. And on the final note on that, really importantly, you've got all the evidence space behind it. But I had the fortune of speaking to someone, a venture capitalist in the diabetes space, and he looked at it and he was like, that is really, really good, that's amazing. But do your algorithms and does your product thing stand up to real world data? I was like, What do you mean, real world data? He said, Well, we've got a data set for 10 years where we follow people longitudinally and over 1.5 million patient days. You can run queries as to whether your algorithms for these people would have produced a proper result. So he said, You've got 24 hours, I'll give you the login. You've got 24 hours to go for it, find out what you can. So a lot of coffee, no sleep for 24 hours, which wasn't probably ideal. Proper robust assessments, all pretty much all of them passed, apart from a couple that we've we've adapted the algorithms because of the results, but it pretty much is robust. So we're going to publish that, and that's what's given us the confidence from an insurance perspective, and they were happy with that to say, well, look, it's an educational thing, it's been validated on the evidence base, it's been validated on real-world data. You're only suggesting it gets you 80% of the way there, and it's clinical things, and that's all over it. So basically, there's nothing like it. And on purpose, we're not trying to make money. What we're trying to do is give it to people with diabetes for free, paid for by the companies who make money off people with diabetes because they're going to want to, and they are doing, giving educational grants to keep their information completely up to date, which means that for everyone, it's always going to be free on a pro level who's a subscriber, and anyone who comes to the website will get three-free queries. But those people who want to take it to the next level and burn tokens like Bilio, like me, 50 quid a month is nothing compared to what you can kind of roll out from there. So I don't care whether people decide to buy it for 50 quid a month or not. I don't really care. What I care about is on the pro level, if you've got 30 questions a month, you can get 30 of the if you had the best educator with you, they couldn't beat what this gives. Because I've I've tried to I've I've challenged myself, I've asked a question, I've already told the answer what I would have done, and the answer that came out is just like it smiles better than me.

SPEAKER_00 19:49

Yeah.

SPEAKER_01 19:49

But it doesn't know the person who's in front of me. Yeah. So it wouldn't be better than me on an on a human basis, but on a knowledge basis, it like it's gonna wipe the floor with anybody. But it doesn't take place of a human.

SPEAKER_00 20:00

So probably all of the questions that I had about G and L Grace, you've already answered them all while you're monologuing. Well, one of them was so what would you say to somebody who is living with diabetes or like a healthcare professional that works within diabetes? I mean, you might have already covered it, but you know, if some people are nervous about the idea of using AI, we always put the disclaimer front, middle, and bottom, like throughout the whole thing, that it doesn't replace your healthcare professional, it doesn't know the person in front of you. But what would you say to somebody who is nervous about the idea of AI?

SPEAKER_01 20:37

Yeah, I mean I I would say I'd say first of all it's well-founded because if you've got an unbounded model that has tapped sources to the internet, it gets information from there and makes a very, very good assessment of getting as close to the truth as possible. But if you've used AI before, have you chat GPT, used um cord, used used loads of things, you do get some rubbish thrown at you at times because within all the information it's got, it's gonna pick up the occasional anomaly. This is different, ours is bounded. We've put everything in there, we have set up the structure and all the trees and how it pulls it together, we've done that. So there's no outside information that comes in. So when the research papers come in, we look at a hundred the best hundred papers in a week, we narrow it down to ten, and then I only put five in. So that only the very best evidence gets in, and it's only me who curates that. So I know from my perspective, the answers it comes out are from a really solid evidence base, from a really solid way of asking questions and how to decipher the information. We've tested it with quite a lot of the um glucose level I subscribers and other people as well. But obviously, it's only gonna get you 80% of the way there. Prescription. It's not deterministic, it's to give you what the average person with your information, so you can put your weight in there, you talk daily dose, etc. etc., and it will give you what the average Person. Don't mistake that for actual advice. It's never advice. It's always on you from a self-discovery learning perspective if you're a person with diabetes. If it smells off, don't go anywhere near it. Like you know when something doesn't feel right to you. Never accept that if it doesn't feel right to you. And if you're a healthcare professional, you could think that this is going to take over your job. And that is the wrong way of looking about it. Because I've done used the explorers and loads of things now and shown pictures and diagrams and like got them to ask some questions when clinic if they've got questions as well. It's a way that you don't have to nerd out on every single algorithm because you want to understand it and you're desperate to understand it. And with you desperate to understand it, all you're thinking in your head is, has this one got a target of this? And which is the one to do this? And which is the one to do this? And when all the while you've got a person in front of you who wants eye contact with you, who wants to have a conversation with you, who wants to see you as a human being, not someone whose like eyes are going rolling in the back of their head because it's too much information. You'll just be able to put it in there, go, those are levers to pull for this system. Brilliant. I want to ask you, what do you want out of your diabetes? Like you come here today, obviously, we want to try and give you as much time in range as possible, but for you, you might not be in that stage of your life. You know, kind of talk to me, let's talk about it. Those are the conversations that make a difference to people with diabetes, not someone who knows every single algorithm. So let's take that bit and give that to a piece of technology that can do it, and then let's use things in clinic that augment the experience for the human being who's on the end of it. So that's the way I would say have a think about it. And obviously you're gonna be scared and not scared, but apprehensive, and you should be, and you should be totally guarded up until you get to a point where you're like, Do you know what? It's just good. Like it's just it's too good not to use, and that's kind of the point that I'm at. Yeah, I'm I'm already there. Like, I've been but I've been.

SPEAKER_00 23:41

I mean, in my head, this is all very minority report to me. All of this, Claude's gonna take over the world. This is what's gonna happen.

SPEAKER_01 23:48

The bottom line is this is coming thick and it's coming fast. Whether you do or whether you don't like it, the world is going to be made simpler by AI, but it's never going to take away the human aspect. So if I now was starting in practice, I wouldn't be as worried about the knowledge side of things. I would be thinking about how can I be the best communicator? How can I be the best person to help this individual in front of me manage their condition? And that takes more than just knowledge. It takes compassion, it takes understanding, it takes time. So that's the skill of the future. The skill of the future isn't knowledge because that can be done for you. The skill is compassion, understanding, and delivery. So that's what I would say if you were a healthcare professional.

SPEAKER_00 24:34

Yeah, and to be fair, I feel like people living with diabetes and also just any of us have all probably encountered healthcare professionals that don't have compassion. Yeah. Yeah, so yeah, I suppose that's a very good thing.

SPEAKER_01 24:47

So all all the all the knowledge robot healthcare professionals will probably disappear in a little while. But the ones who really like human beings will stay around. So that's what can only be a good thing.

SPEAKER_00 24:57

So, in terms of what's happened between like January and now we're in April, it like everything that's happened in those past few months, is is this what you expected to happen or has it all been very I I I re- I re-listened to this. From from like March, maybe ATTD, like probably from that point it's all kind of It's just gone bananas.

SPEAKER_01 25:23

So I I would say no. I would say I listened to the podcast that we did the last one, the quarterly. I had quite a bad time personally in January.

unknown 25:31

Yeah.

SPEAKER_01 25:32

Really not a good time personally, and like took a massive nosedive, but then came out of that about mid-February, just before ATTD. And then went ATTD, and then it's just all happened. There's a lot of conversations that happened there that made me think differently about how education should be delivered. Some stuff on digital twins that completely blew my mind, which we'll probably talk about in a second. But it got me to thinking about what's possible, and then obviously came back, and then it just so happened that Phil introduced me to Claude and Claude Code, and then he said, Look, we can do this, we can do it if you want. And then I have got a fairly obsessive personality, quite obviously. No, really, and it yeah, I just basically took took took the bull by the horns and uh turned 120 pages into something which is interactive, fun, easy accessible, and then G and L Grace has just taken it to another level. I would be amazed if people who try it are just like, wow, how have you managed to get that? And the answer is it's taken 20 years of learning and citations and publications and publishing on structured education programmes and speaking at conferences and whatever to get the clinical knowledge base and application, but combine that with your best mate and you're willing to work together and go like bilio on it and produce something which is unbelievable. Yet it's only possible with 20 years of AI and developing experience plus 20 years of clinical experience coming together of two people who get each other. It's just a rare, a rare, rare thing. So I will be surprised if people don't love it, to be honest. And if they don't, I don't care, I love it. So like if nothing else is made for art it's made for me, I'll be using it. It was it's I don't like doing boring repetitive tasks. I hate doing boring repetitive tasks, and this stops that happening. This allows me to be creative free. Yeah. And the the creative stuff you can like you make for it, it's it's bonkers.

SPEAKER_00 27:25

Yeah. Cool. Well, uh yeah, like you said, it was probably just uh like a once-in-a-lifetime thing. It was a very like freak occurrence that it could happen. And then, yeah, some opportunities came along, and then you probably just couldn't.

SPEAKER_01 27:39

Well, the the thing of like getting access to the 1.5 million patient day is like that this boat just basically came across and he'd looked at the he'd looked at the alcohol and drugs guide that we did, and it was like, I've never seen anyone be as open and honest and upfront about what it's really like with type 1 diabetes and the things you need to take into consideration. As a favour to you for doing that, because that's going to help loads of people. You've got 24 hours.

SPEAKER_00 28:02

It sounds like the Tony's talk of diabetes.

SPEAKER_01 28:06

You've got 24 hours to go on this playground and get your model because the the amount of tokens that I will have burnt doing those. I don't know how much I owe him. It's probably quite a bit of money, I should imagine. But we we value the cell that you've given me, right? No, we well we did no, but he's like, he's let us do it. Let us do it for free. He's taken the hit because of the the good work that we do, which is unbelievable, obviously. And again, just a freak occurrence at the right time. But at the same time. I thought what I was gonna say then. It was something profound as well, and it's just disappeared.

SPEAKER_00 28:38

Sorry, that's me. That's me interrupting you. Are you sure?

SPEAKER_01 28:42

My brain at the moment, there's too many tabs open. There's too many tabs open.

SPEAKER_00 28:45

There are, there are.

SPEAKER_01 28:46

No, I know what I was gonna say. It just sort of like it was getting the mixture of learning how to do cloud code with Phil, and then obviously meeting this bloke, and he was like, Yeah, you know, kind of you can in 24 hours basically validate your algorithms and your approach and G and L grace on 1.5 million days if you know how to do it, off you go. And I was like, I'm I'm game for that challenge, I know how to do that. Like, and that's the beauty of being having research papers behind you and knowing what to do. So we we did those analysis, and it's like I say, we've already written up the paper, we're about to stick it in for review. Um, and because I know what a lot of people say is oh that that's like black box stuff, you know, kind of you haven't got anything behind it. I'm telling you, there is a publication coming and it is ridiculous. And it was written, written half by Grace. By G well, all the diagrams are written by by Grace. The the main bit was written by me and then improved by her and then checked by me, and then the seven diagrams that go with it they just blew you away. Like I can't believe I it would have taken me two months to make those diagrams. Two months twenty minutes.

SPEAKER_00 29:53

Yeah.

SPEAKER_01 29:53

Blew them out of the water in twenty minutes. It's unbelievable, and it just makes all the difference.

SPEAKER_00 29:58

I suppose that kind of leads me on to the next section. So reflecting what's happened so far and thinking about what's gonna happen next. So would you say that that's probably the thing that surprised you the most?

SPEAKER_01 30:09

Oh, without a doubt. I mean the the explorers were fun, but I knew what the answers were to explorers because I developed the algorithms. So I did I made the exercise one based off the ISPAD guidance in 2022, all the structured tree questions. So they obviously I knew what the answers were gonna be. So and that that's nice because it makes a plan really easy to make if you're a healthcare professional uh for a patient and adjust it as you see fit. And if you're a person with diabetes, it'll get you 80% of the way you can discuss with your healthcare professionals. So I kind of like I had that in my mind for ages, and I've I've already done the PDFs. You send the PDFs that we use at the hospital, but those are PDFs that are sat on a crusty old thing somewhere. These are like live, three clicks, and you're in, and it just saves you so much time. So I thought, yeah, that's brilliant. But then the grace thing, I was not expecting number one, just the quality of the basic questions. But then when I put those manuscripts in and what came out when I say it blew me away, it's like no exaggeration. Like the paper that we have done that's gonna go into nature AI journal or whatever.

SPEAKER_00 31:11

Yeah, yeah.

SPEAKER_01 31:12

I'd be amazed if it doesn't get accepted, and also if it doesn't, it'll be getting accepted somewhere. And it is just it's just a mad, mad piece of work.

SPEAKER_00 31:21

Yeah, so watch his face. We'll let you know when that's out. If it comes out, it will. Yeah. So looking back, I mean, you could probably there'd probably be a million scenarios for this. So is there anything that you thought would work but didn't work?

SPEAKER_01 31:35

Oh my god, loads of stuff I'm loads. As as. And that's the beauty, that's the beauty of this. And I think the the like the people who subscribe to the the blog, and the more people that will s will subscribe to once they want to use uh Grace Pro, hopefully, you know, just get your 30 queries done. You are gonna get emails from me that'll probably like for example, when we're setting up the registration for the email and password, we thought we'd got it sorted, we sent it out, and it's like, oh my god, I already had like 50 emails coming back going, I can't log in because the password reset doesn't work. It's like so we thought we fixed it again. So then we sent out thinking, oh yeah, we sit fixed it, fair enough. And then it hadn't fixed it. I was like, so it that's just my eagerness to push things out. But at the same time And you won't stop. We try and stop it, but that's the problem. We try like I'm not gonna push out anything that's clinically unsafe. But when you think you've done something, I'm not gonna wait three or four days to get a verification when I could have been doing those three or four days doing something else. If it's low risk, like a few blogins, yes, it might have pissed a few people off. But that's just the downside of uh something that moves fast, something that moves really fast and gets things like GNL grace done. There's gonna be small things along the way that annoy you. So if I do annoy you with the occasional email, I am sorry about that, but that's just part and parcel. You don't get GNL grace if you don't have a bit of that. So that's you know, it doesn't surprise me because I've that's been me my whole life. But uh sometimes I still think I wish I wouldn't have sent that, and then I just look back and I go, well, it's just part of who I am. So yeah. I've been stopped by obviously you stopped me doing the disastrous stuff, but um the the small stuff that's the yeah you tried the small stuff that you you can't be privy to that happens at three o'clock in the morning or something, like you can't do it.

SPEAKER_00 33:14

Yeah, that's because usual normal people are sleeping at that point, not sending stupid voice notes at silly o'clock in the morning, saying I broke the website or whatever it was. I mean, there's probably a lot more of that to come. So is there anything that you feel like is kind of unfinished from this quarter, or do you feel like this chapter's done, or do you feel it's just a work in progress?

SPEAKER_01 33:36

You get G ⁇ L Grace out properly pro and max on on Wednesday, so everyone will receive an email on Wednesday saying, right, you guys can log in and go and have fun. We would love some feedback. So the beauty of this is it's an evolving beast. How it's set up now is not how it will look in a year's time. If you want it to be able to do, I don't know, you want a button on Grace Pro that is something like explain this for a five-year-old, explain this for a 12-year-old, explain this for an 18-year-old. You want that button because you deal in education for those areas, tell us and we'll put it on there for you, and then you'll click and that will do that mode. If you want it to produce something for you which is kick-ass and wicked, like a poster for diabetes day or something like that, tell us and we'll like we'll put it on there. It's no problem. So the more feedback you give us, if you put, oh, do you know what the evidence base is missing on this? Tell us because we can go out and Grace has this structure where she's got these concepts, and at the moment she's got 60 concepts, all to do with like alcohol, ID systems, CGM. She's got all these trees. We can add more trees in, we can put as many trees as we want in there because the base layer and the questioning and the and the logic that she uses, and she also uses a via negativa approach, which is basically major in the majors, and if it does not on a major signal, it doesn't get discussed because there's too many things in diabetes to throw every little scenario in. Major in the majors, and that is the biggest thing um about the approach with it.

SPEAKER_00 34:57

You mentioned getting feedback from users as well just then. So obviously, you look like glucose never lies have thrown up on you because you've got it on your head or on your t-shirt, you've got it behind you. So tell us a bit more about that.

SPEAKER_01 35:11

Yeah, yeah. So um the first hat is on the way to two uh things where you can see we've got a raft of hats here, if you can see on the thing. We've got white ones and we've got black ones. We figured hats are the best way because they're quite cool, the logo looks decent, and you don't have to worry about sizing. So for us, it's just easy to ship out. So people who give us really good feedback are gonna get a hat. So Ali Boulderstone gave us an amazing feedback on the explorers, and that's on its way out to them. And then Frederick was another one, he's the guy who gave us the access to things. Yeah, so we are base we are so we are basically gonna give hats to the best feedback that we get. Yeah, and we'll continue to do that. And if you're at Diabetes UK, um if you email me with pictures of answers that Grace gets to you, I'll give you a hat there. So if you're a healthcare professional that's going to Diabetes UK next week and you use GNL Grace and you get some wicked pictures and you email them to me and come and find me at Diabetes UK. I want to have these hats with me so you can have one if you want. Cool. So I'm gonna have to round it up. So I've just been told I've got to get out now because it's my mother-in-law's birthday, so I'm gonna get in a massive amount of trouble. So I'm just gonna say you're gonna get another podcast on Wednesday morning. GL Grace is gonna go live. It's gonna blow your mind. Just as simple as that.

SPEAKER_00 36:24

Yep. Thanks, John. Let's take a wrap.