#1986 Trio Version 1.0: Open-Source AID
Trio Version 1.0: Open-Source AID
Dan leads development on Trio. It now runs Omnipod 5, and he explains how an unpaid team scattered across the world keeps beating the industry to it.
Jump to a moment
- Trio 1.0 reverse-engineered the Omnipod 5 pod communication protocol, so the pods now run in open source AID the way Dash pods always did. The point is not the pod — it is that you are no longer locked to whichever CGMs Insulet has partnered with. Pumps supported now include Omnipod Eros, Dash and 5, several MiniMed models, Dana i and RS, and the Medtrum Nano, with the Tandem Mobi in closed beta. Sensors include every Dexcom, Libre 2 and 2 Plus, Libre 3 and 3 Plus, Eversense E3 and 365, and the Accu-Chek SmartGuide.
- The release itself was a year of work by more than twenty people — 112 new features and fixes across more than 1,500 commits. The algorithm was also ported into Swift so it now runs in the iPhone's own language instead of the one it was written in a decade ago, and that port is becoming a peer-reviewed paper with people at Stanford and UC Davis on how to safely update an automated insulin delivery algorithm. Dan's framing: "We're not rogue cowboys."
- The clearest picture of how this community works came from a community Zoom call. A user said he had a visual impairment and asked whether the lock-screen display could be bigger and colored. Dan went to bed in Germany at 1AM. By the time he woke up, a developer in the US had the feature built and opened for testing.
- Dan's answer on AI is worth the whole segment, because he builds with it and still draws a hard line. It helps with interface work, log analysis, writing tests, reviewing features — and never with the algorithm or with dosing. He and Tim Street sent the same dozen food photos to the models roughly 27,000 times, about 500 to 600 runs per image. Dan's paella in Barcelona, which he bolused for 60 carbs, came back from GPT at 423. The failure he keeps pointing at is that you cannot see the variance: you get one confident answer, with no way to know whether it landed in the middle or at the far edge.
- His structural complaint is data ownership: "we are the currency." A Clarity report of your last ninety days is not access to your raw data, and almost no manufacturer gives you the latter. His analogy — imagine Gmail would only ever hand you a generated PDF of an email, never the email. It also has a practical consequence: the only reason users were able to donate data toward Loop's FDA submission and the European OPEN project is that running open source gave them access to it in the first place.
- Trio on GitHub — The code, the onboarding guide, and the browser build Scott uses for Arden
- Nightscout Foundation — The nonprofit that funds the hackathon, developer devices and travel — donate to the annual hackathon here
- Why the Cheese Sandwich Is the Downfall of AI Carb Counting — Tim Street's write-up of the toast no model and no paid app has ever gotten right
- Episode 1243: DIY Insulin Algorithm AIAPS — Dan's first time on the podcast, if you want the longer version of his own story
- MiniMed Flex — MiniMed's smallest pump yet, the size of two stacked vials, controlled from an app
- Contour Next Gen — The meter Arden carries — second-chance sampling and a downloadable coupon at the link
Every word of the conversation
Welcome and Sponsors0:03
Friends, we're all back together for another episode of the juice box podcast. Hey. Nothing you hear on the juice box podcast should be considered advice, medical or otherwise. Always consult a physician before making any changes to your health care plan. The holidays are upon us.
They're coming quickly, and you would like to feel like you're giving back, but you don't have time, but you wanna help. T1dexchange.org/juicebox. Head there now. Complete the survey. Trust me.
You do that. You've helped. T1dexchange.org/juicebox. 18 or over, type one diabetes, either for you as the adult or as the caregiver. US resident does not take that long to sign up and do the whole thing with the survey.
Trust me. You look into the t one d exchange. They are a big part of why things get done. You could be a big part of why the next thing gets done. T1dexchange.org/juicebox.
This episode of the podcast is sponsored by MiniMed and the Contour NextGen blood glucose meter. MiniMed, of course, the innovators behind the MiniMed Flex system, a small, sleek, and discreet insulin automated delivery system controlled from an app. Nothing comes close to freedom you can feel. To learn more, visit minimed.com/juicebox. There is one blood glucose meter in this house, the Kontoor Next Gen blood glucose meter.
That meter is on my daughter's person right now and at all times. It is incredibly accurate and waiting for you at kontoornext.com/juicebox. Holidays are coming fast. Cozyearth.com. Use the offer code juice box at checkout to save big on everything in your cart.
Go to cozyearth.com. Use the offer code juice box at checkout, and you will save.
Meet Dan1:49
Hi. My name is Dan. I'm a 33 year old type one diabetic from Germany. I'm in... Engaged in the open source AID community.
I've had type one diabetes for the better extent of almost three years now. And, Scott and I happen to know each other because I am the lead developer of the app Trio, and that's why I'm here today.
And I am excited to talk about Trio and find out what you guys have been doing with it and where you think it's going. So we have to give some context for people who are just like, I don't know what you're saying. So I usually dumb it way down and you can fill in the the rest. But Go ahead. Shoot.
Here's how I think of it. I think of it if there's a community that's connected through, you know, being online together who has, in the past, created a do it yourself algorithm that helps people with their insulin. And over the years, those apps have been modified, changed, improved. Sometimes they branched off into other apps, and there's... I don't know how many tens of thousands of people in the world probably using the loop or Android APS or Trio.
How how... Do you do you have an idea for how many it might be?
The dark number is pretty high actually, about, like, a hundred and hundred and fifty thousand people. The people that, like, we have numbers for is probably half of that. Mhmm. But it's it's a pretty large number, actually.
That's great. And and it's it's fantastic, honestly. I I would always wanna say that Arden Arden used Loop the first time when she was in high school because a listener of the podcast came on one day and said, I just... I really wish you would do this and learn more about it so you could talk about it. And I thought, oh, okay.
Well, I'll see if Arden's interested in that. And, you know, besides the stint with Omnipod five, Arden's been using either Loop or now Trio for years. And Trio... It's funny. Like, I'm...
I think if people have heard enough of these conversations, they know that technically, don't I don't... I'm not the right person to ask about these things. I don't really understand or know know how they work. I just know how well they work and what a good experience we've had with them. And when Mike came along and told me about Trio...
Because I think at that point, we were using Loop and he was like, hey. You should try this thing we're working on called Trio. And I said, cool. And then I just switched because people I knew and liked and trusted said to, and here I am. So...
Yeah.
Yeah. How much of that do you think people don't understand now? Like, is that pretty clear?
What Open Source AID Actually Is4:21
I hope it's pretty clear. I mean, I can I can I can try to maybe say it in a couple different So you basically already, like, summarized? So open source automated delivery insulin... Sorry. Automated insulin delivery apps.
There are a couple out there. They're totally free. They're community built. They're mostly built by diabetics and for people with diabetes. And they essentially allow you to connect your insulin pump and a compatible CGM to your phone.
And then they basically talk to each other via the phone. And on the phone, there's an algorithm running that is adjusting insulin based on your blood sugar and on various things that the algorithm will, you know, calculate and compute. And because it's open source, anyone can basically build it and they can run it themselves instead of having to go to any company and buy an expensive pump system other than, you know, having to have the hardware, but you don't have to pay them for access to an algorithm. And, yeah, it's it's been great. It's been a journey that's been going on for the better part of a decade, actually.
So the the first system that, you know, was an open source AID came out in the in the early twenty tens. And now we're in 2026, and people are still using this, and people are still innovating, and we're still pushing the boundaries. And I hope we're also still pushing industry Yeah. To to do a little better for us.
Well, I've recently said on the podcast, and I'm happy to say it again here where maybe more people will hear it in context, I genuinely believe that the work that all of you do has a lot to do with how the industry does get pushed and what motivates them to keep moving. Because, you know, it would be easy to be a big company and make a pump that works and just go there. It works. You know? We'll go do something else now or let's just start making profit off of that.
But I think there's pressure that comes when you can download something that works as well as Trio does. So Yep. Yeah. I I appreciate all the effort, actually. I greatly...
I I wish was... I actually found myself telling somebody recently, I should probably not say this out loud, Dennis, because somebody's just gonna steal my idea. But I... I've... I really...
I've had the idea recently that I would like to I'd like to get together some, like, OG people and some people and and some people from along the way through this whole process and do... I I don't wanna call it, like, maybe a series of conversations with people who have been involved with it over the years just so, you know, not not different than a documentary, really. Like, just hear people's stories about, like, over the years, how it started, you know, people that jumped in and helped along the way fell out, new people that came on. I just think it's fascinating.
It it is a it is a super fascinating story. I mean, it's it's... The the thing is it's like it's like this ever growing tree. Right? And so, like, you know, some leaves fall off and then other leaves grow again, and this tree is still standing and it's growing and it's, you know, getting taller and taller.
And it's it's it's a bit of a of a, you know, romanticized image I'm kind of trying to to paint here, But it's a bit like that, honestly. And it... It's it's so interesting to meet people from across the world and, you know, when when some people pop up. Just recently, for example, there... There's this young guy out of The Netherlands.
He kind of just popped up on Discord, which which, you know, for people that don't know what Discord is, it's it's a it's a chat thing a lot of gamers use. And part of the open source community is organized in Discord. And he just popped up, he was like, hey, guys. You know, I'm using this. This is super cool.
And I heard, you know, you might be looking for help, and I I have I have cool skills, and I could maybe help here. And he's essentially, like, working on a couple CGM integrations now with the whole team, and he joined the community just a couple months ago, and now he's doing that. And it's... That's the beauty of the open source AID community. That's fantastic.
Just we just do... Sorry. No. I don't wanna swear here, but, like, you know, it's it's... We we just we just see see things that we wanna do, we see things that people need and we just we just move and we just we just not waiting for for industry to to, like, do stuff and bring stuff forward that we need.
Yeah. We just basically do it ourselves. Yeah.
It's funny. I... It it reaches the people you don't know. It... It's...
You would never... I mean, I spent a lot of my summer going from event to event with, John from SugarPixel. And at one at one point, I opened my phone. It was, you know, really early in the morning. I knew Arden was still asleep.
I heard her blood sugar was a little high. I didn't see any correction that I wanted to see, and I gave her, a little bolus. And I kinda watched John watch me for a second, and then it hit me out of nowhere. I'm like, I'm using an app he made. Like like...
Right. And and he's just one per... Like, he's one, like, random person. And then there's... And then I...
You know, you guys come over at at ADA, and I was like, oh, there's three people standing in front of me that are pushing this whole thing forward. I actually said to somebody at ADA, you three walked away. And I said, those three guys are propelling this entire industry. And and they're gonna walk right through this room, the average person wouldn't know who they were. Contournext.com/juicebox.
That's the link you'll use to find out more about the Contour NextGen blood glucose meter. Once you get to that page, just head up to the top and you can click right on blood glucose monitoring. I can even do it with you right now. You just go to meters, click on any of the meters at all. I'm gonna click on the Nextgen, and you're going to get more information.
It's easy to use and highly accurate. SmartLight provides a simple understanding of your blood glucose levels. And of course, with second chance sampling technology, you can save money with fewer wasted test strips. As if all that wasn't enough, the Kontoor Next Gen also has a compatible app for an easy way to share and see your blood glucose results. Contournext.com/juicebox.
When you get to my link, just scroll down and you're going to see things like a buy now button. You could register your meter after you purchase it or even... Oh, what's this? Download a coupon. Oh, and receive a contour next.
Oh, there might be a free meter on this page. You have to go check it out. Contournext.com/juicebox. That's contournext.com/juicebox. Head over there now and get the same accurate and reliable meter that we use.
If you're a parent of a child with diabetes, you know nighttime can be one of the hardest parts of the day. You go to bed wondering where their blood sugar is headed. You wake up to check numbers, wondering if everything is going to stay steady until morning. The new MiniMed Flex system from MiniMed uses the same trusted automation people already know, now in MiniMed's smallest pump yet, and fully controlled from your phone. And here's something that really stood out to me.
People using recommended settings achieve 91% overnight time in range. Think about that for a second. Not because it's a statistic, because every hour they stay in range is one less thing to worry about while you're trying to sleep. Learn more at minimed.com/juicebox. You'll find links to MiniMed, Contour, and all of my sponsors at juiceboxpodcast.com and in the show notes of the podcast player you're listening in right now.
By using those links, you are supporting the production of the podcast and helping to keep it free for everyone. It's really fair. I just... I mean it. It's it's fantastic and and all the other people that have been involved over the years.
Yeah. So so anyway, tell me, I guess, like, just for a tiny bit of context. Yes. Loop was... It was a...
Ivan, and the Branch That Became Trio11:48
You know, Loop was going for a while. And then it it branched off. I'm gonna use terminology. I don't use it my whole life. It branched off, which means what?
You guys took the code and then...
Oh, okay.
So Yeah.
So in in in... I mean, defense is the wrong word here. No. But in all honesty, LUB is still kicking and LUB is still alive and LUB is still doing well. Mhmm.
So Loob is still there. There was a there was a developer, a Russian guy called Ivan Valkov. And in 2021, he started introducing a couple features to Loob, and they just didn't really get into the the Loop code base, so they just weren't added to Loop. And, you know, being being an open source guy that he was, he was just like, hey. You know what?
I can just do my own thing here. And, that branched off, and he he made up an app, and that app used to... Like, got to be, what then was called FreeAPSX. And, what he did, he was taking the algorithm, the OpenAPS algorithm or the ORF algorithm. So that is an algorithm that was developed in the twenty tens.
And some people here might know that it's also the algorithm that is running in Android APS, so an Android based AID system. And he built up this app. And then over the years, you know, a couple people came on, a couple people went off. And in in late twenty twenty two or mid twenty twenty three, a large group of people branched off from from a project that was... That is also still existing, and we came up with Trio.
And then ever since then, we've been developing Trio, and we work very tightly together actually with the folks that are doing loop. We are well connected with the... With our friends over at Tidepool because, you know, the people at Tidepool, part of what they can do is also, you know, push the open source envelope as well. And the Trio team is about, you know, a a dozen core developers. The wider team is about three dozen people, and we have a community of a couple thousand people that are using Trio every day.
And, yeah, that's that's... I guess, that's that's the Trio story here in in thirty seconds.
I have to ask, is Ivan still involved?
No. Unfortunately, Ivan had to kind of, like... Well, I guess, he he could no longer really participate when the when the Ukrainian Russian war broke out, when Russia invaded some parts of Ukraine, and all the sanctions were started to, you know, politically and economically stop that Yeah. Or, like, stop Russia from doing that, he could no longer, like, participate in in development, basically, with the West, I guess. And so he had...
He kind of pulled back. And to my knowledge, he's still actually developing some sort of AID up in Russia for Russian people with diabetes.
Good friend.
But he's... He he... Like, there's a big telegram group going wherein they have a community going, but he's no longer involved in in Trio or in in anything that is, like, loop Trio, Tiepool So the West. The Europe.
Yeah. The reason I asked is because I I... Listen. Again, there's a ton of people involved. I couldn't possibly know everyone's names.
But I once heard Ivan from somebody. There's, a guy in Russia. His name's... And so whenever I I reach for a generic person involved in, you know, AID DIY, I always say, there's a guy in Russia named Ivan. He's doing the...
You know, he's doing it. And then one day, somebody's like, I think he's the, like, part of the trio. I was like, oh, how random is it that that's the name I chose? And it was just the one that stuck in my head. I hope he knows how grateful I am and how many times I say his name even though I have no idea who he is.
Yeah. But
it... It's just I think that's a fascinating look at how it gets... Like, said leaves fall off, leaves come on, but almost how it's, like, passed, like, a baton, you you know,
from It's it's kinda like passing a baton. I mean, so so in in in, like, in in all respect, Ivan was never involved in, like, what became Trio. Okay. Like... So we are basically...
You know, we're always building on the... Like, we're always standing on the shoulders of giants. Right? So, basically, Ivan was one of the giants that we're standing on now. We are the...
If if you wanna call that, we are, like, the third generation of a project that Ivan started in 2020, 2021. And, it's also, I guess, to some extent, the reason why we're called Trio. I mean, the the origin story of the name is a bit, you know, you need you need glucose, you need insulin, and you have the app that controls it. That's a Trio. You know?
You have a CGM. You have a pump. You have the phone. That's a Trio. But but also, we're the third generation of, the iOS running...
Oh, sorry. The algorithm that is running on iOS and... Because we evolved out of that project. And so... Yeah.
I'm... I've I've never met Ivan. I've never talked to Ivan. A couple of people that are still, you know, some of the OG Trio folks, they actually knew him. My...
Myself, I I didn't. I I read a a ton of his code. I've worked through a ton of his code. I've I've read a bunch of the things that he's done and and and he's still doing. But...
Yeah. And that's Fantastic.
Basically No. That's fantastic. Well, let's let's dig into, let's dig into how much you're gonna save when you go to cozyearth.com and use the offer code juice box at checkout. Bedding, blankets, pajamas, robes, towels. There's too much stuff.
I can't just list everything. It's crazy. I'm telling you for men, for women, for the bath, for the bed, you want cozyearth.com, and you also don't wanna forget to use the offer code juice box at checkout to save on your entire order. Not the most expensive thing in the card or one thing or just one thing that's on, like, in a list. So everything in your cart, you're saving.
Cozyearth.com. Use the offer code JuiceBox. You wanna be cozy? This is how you're gonna do it. This is gonna give you the quickest story.
I got my Cozy Earth blanket on the sofa. I was, trying to watch a baseball game last night. I threw the blanket on. The next thing I know, son's tapped me on the shoulder. He goes, hey, man.
You're asleep. Do you wanna go to bed? Should I shut this off? So comfortable. Everything they make feels like that to me.
Everything from the towels to the sheets, from the blankets to the pants. Just go look. Cozyearth.com. Use the offer code juice box. You guys have just recently put up an update to Trio.
So what got changed, and why why would would people be excited? I have a note in front of me here that says, I'll be updating Arden's Trio soon. So that's from the person who handles handles it for me. So...
Trio 1.0 Runs Omnipod 518:09
So so I I I guess one of the one of the, like, most impressive and really cool things that we were able to do with Trio is that Trio one dot o, so the the grand release, is supporting Omnipod five. Oh. So what what does that mean? Because you're like, wait. What?
Isn't that like a commercial system and wait. What? So we were able to basically reverse engineer the the Omnipod five pods, so the communication protocols. And we're now able to talk to the Omnicod five pods just as if they were dashboards, for example. And so you now can use them with Trio and our algorithm.
And I don't I don't wanna downtalk or trash talk in Slip by any means, but a lot of people are not really happy with the o p five algorithm. And so Trio one .o... And also for that matter, in all fairness, also the the most recent loop version, you can use the o p five pods in open source AD. And so the really, really, really cool thing about this, and I I really wanna emphasize on that, is it unlocks you from only being able to use the CGMs that, you know, Insulet has partnered with. So for example, if you're in The US and if you're running an Eversense e three sixty five CGM, you can now use it with an o p five pod and you can use it in Trio.
Or if you're living in, you know, in Slovakia and you're using some some older Libre sensor that isn't even sold in in The US, for example, a Libre two plus, you can use it with Trio. Or if you, you know, if you wanna use some kind of CGM that is maybe only sold in Europe. And for example, Roche, the the the company here in Europe, they have a new CGM. It's called the Aquachex SmartGuide. And we've also added support for the Aquachex SmartGuide with the recent release.
And so we basically unlocked that thing for all of our users, and we are offering true interoperability for for people that wanna use Trio. And before I forget it here, we also added direct support for the Abbott Freestyle Libre three and three plus. So you now not only have o p five, you also have, you know, all the Dexcom sensors, and you also have the latest Abbott sensor. So, essentially, you can loop with almost all the pumps in The US and all the CGMs in The US.
Yeah. It's funny how you say loop because it's it's it's a ubiquitous word at this point, isn't it? Like...
Yeah. I mean, to be fair, I mean, DIY looping is is is basically the the generic term for what we're doing here. And even the commercial systems, they are looping systems. Mhmm. And I guess the the guys that, you know, at Tidepool, they just came off this really good name.
Well... And so, like... I mean, right now, there's what? There's tide pool loop in the twist pump. Yes.
MiniMed is got their new flex pump out. Tandem's got the Mobi. They have the the control plus. They're they're gonna... They're, I mean, they're getting ready to introduce a tubeless pump, aren't they?
And isn't isn't mini med too? Like, there's a lot of pumps coming.
Yeah. Yeah. There there... There's a tubeless pump by mini med also coming. They they...
I think they've talked about it in in or around ADA, and I've also seen them on on on some podcasts talking about it. Beta Bionics is bringing out the the Islet Mint. Right? So the... They have also have a patch pump coming now.
And then, yeah, as we're saying, MiniMed has the flex, and Tandem is also going to release the the patch Mobi version. Mhmm. Not just, you know, the adhesive with the... Where you can slide the Mobi in, but an actual patch pump version of it. And speaking of Moby, actually, I mean, you were at ADA, and we talked about this basically right when it happened.
But at ADA, we we not only announced that, you know, by the end of the summer, o p five would be loopable with Trio and Loop. We're also working on the Tandem Mobi. The Tandem Mobi is, unfortunately, I wanna say, unfortunately in quotes, still in close beta testing, but we're still working on it. And, you know, fingers crossed that it'll land this year for, you know, public use. And then also the 10 demo, we will become loopable with a future version of Trio and Loop.
Yeah. Do you ever... Do you have any feeling for how... I mean, maybe it's not even something you can share. I'm not sure.
But I... I'm trying to put myself in the I'm trying to put myself in the position of somebody listening who's hearing. So I download some code from the Internet, and then it attaches itself somehow to my phone. And then it's talking to my insulin pump, and it's not using the algorithm that the people that made the insulin pump want me to use. It all probably sounds fairly, like, odd or it could sound scary to people who don't know what they're they're doing.
But my question is is do you know what the companies think about it?
What the Pump Companies Actually Think22:57
I mean, it it... I guess it depends it depends on the company, and it it depends on who in the company you would ask. Right? I mean, some some people, they are really not happy. Obviously, when they're working in, you know, regulatory or compliance, they might be scared that something can...
Could come up. What I can say is that that, open source AID is perfectly perfectly safe. It's it's not just like, you know, some shady guy sitting in in some garage somewhere and, you know, doing some weird... This is this is built, like, by actual software engineers, and there goes a ton of testing into this. And we...
Like, this this is very, very safe. And talking about the the commercial companies, you would actually be amazed, like, walking through ADA, for example, if you walk through the convention and, you know, how many people that work for some of the commercial companies are also, like, using Trio and Loop. You would be amazed. I mean, I I cannot name names, and I cannot name company names. But you walk up to, like, a couple of the booths, and and, you know, you also know a couple of faces when you go to these conferences.
And they go like, oh, hey. The Trio crew is here, and then you just start start chatting away. And then it's it's pretty interesting when you're standing basically at that booth, big logo behind them, and not... They're not even using that system. And so it just it just speaks to to the safety and to the to the outcomes that you can get with the systems.
Porting the Algorithm, and a Paper Coming24:19
Yeah. Speaking of safety, actually, so one of the one of the main things that also happened in the in the Trio one that I released was that we... Essentially, we ported the algorithm that is running in Trio, and we ported it into another language. So for for lack of a better description, so the the algorithm in Trio basically speaks the native language language of your iPhone now. It it was speaking a different language, and now it speaks the language of your iPhone.
It was speaking this different language because it's it's the algorithm that that was built in the in the mid twenty tens. And we not only ported that algorithm. We actually, like... We were a bigger team, and we we were working together with folks out of Stanford and folks out of UC Davis, and we actually are going to publish a paper on this. So there's a scientific research paper coming out of the algorithm port showing how to safely update automated insulin delivery algorithms.
And, you know, we were doing that together with a couple people at UC Davis. A very young gentleman is the is the head author of that paper. He's also type one himself. And, you know, it's it's not like a couple people sitting somewhere doing some weird stuff. It's...
We are actually approaching this from, like, a very, very professional and safety minded angle. Yeah. And... Yeah. It's
Well, I mean, for example
I I could talk about this for hours.
No. I'm I'm happy to talk about it for hours. So I... But I'm, like, you as an example. Like, you have a job somewhere.
Right? And Yes. And I'm imagining you're a data engineer. I don't know what you call yourself or something to that effect. Right?
You're coding or
I'm I'm I'm a software engineer.
Software engineer.
I'm a software engineer and a and a product manager. Yes.
And and so it's a thing you do professionally. You're surrounded by other people who have the same story. I've... I've... Listen.
I've heard people who work at those companies who have also been involved in stuff like that off and on, or the company is sometimes hiring people who have been involved in, you know, vice versa. So there's... It's a passing of... There's a lot of talent out there, I guess, is what I'm saying. And, and a lot of you are putting it in...
Into good use for this. I have a question of, like, is it all always remote, or do you ever get together? And as a group, how do you... Like, isn't it more productive to be together, or does that not really matter?
A Hackathon in Vancouver26:35
It it it does matter, actually, and it's so funny that you mentioned it. I I was thinking of how can how can I bring this up, and you brought it up? It's perfect. Thank you thank you for that. So we we do meet at at scientific conferences.
We we try and meet up at these at these conferences. But three years ago, we... The the NICE the NYSCAA Foundation, I should say. So the nonprofit that is funding a lot of this work is... Has been doing a a hackathon.
So hackathon is basically a multi day event where you where you meet, where you work on problems, you know, where you come up with cool solutions, and a lot of it is is around code. And, we're actually gonna, have another hackathon, coming up in Vancouver, this year. And there was also one in Vancouver last year, and the one before that was in San Diego in 2024. And, yes, we are meeting there. I mean, there's, like, 70, 80 people going to be there this year, hopefully.
And
I don't know if
I if I can actually plug this. Scott, please, like, you know, interrupt me if I can't. But this is this is completely, like, donation based. Right? So, you know, this is a nonprofit, and there is, like, no for profit idea behind this.
And then we're still, like, raising funds, actually. And we're still about, like, 30 away from our, like, funding goal for this year's hackathon. So, you know, if if anyone wants to donate, it's it's pretty easy. It's donate.hackdiabetes.io. You can just go there.
And as much as you know, a Starbucks coffee, like, $5 can make a difference. And yes, we're gonna meet there. It's gonna be 80 people from across the world, out of New Zealand, out of Australia, from from Europe... From many countries in Europe, actually, from The US, from Canada, and we're gonna meet there for a whole weekend, and we're gonna sit there and collaborate. But to kind of answer your question as well, a lot of it is actually remotely.
And that's kind of the beauty about this community because whenever someone, like a a regular user... Like, whenever Arden would have a problem, she can come up on, a Facebook group, and there will always be someone up that can help her
Yeah.
And, you know, twenty four seven, basically.
I I struggle with this sometimes because the people who help manage my Facebook group, I'll do it as a... You know, they're just donating their time. And I I always wish I could help them sometimes because they are so valuable and move people to the right information and and help them to get clarity when they need it. I'm on this webpage now. So it's donate.hackdiabetesdot...
Say it again? Io.io. Okay. So then...
And and it'll it'll forward you it'll forward you to to the NICE COURT Foundation website, and then there you can donate.
Oh, yeah. Look at that. So there's different levels. You could donate... Asks for another amount of $15, 5,000.
Scott Donates On Air29:25
You can put in whatever you want. Yeah. What do I do? I'm gonna put money into this. How do I do...
And you guys should... If you... Anyway, I can't tell you what to do with your money, but I would do it if I was you. Well, here, there's a there's a $250 level here. I'm gonna do two of those.
That is that is super cool.
Thank you very much. No. Are you kidding me? You guys... I I I couldn't...
If I could send you a million dollars, I don't know that it would be enough. You know what I mean? So... Alright. Here we go.
I can do this. This is simple. I'll put my name here.
I'm doing it right now. It puts in my address. There there is also there is also, like, a little store there. You can buy cool T shirts. And then, also, if you, you know, if you're open to supporting our community on on a regular basis, there's something called a subscription, basically.
So you could donate donate, like, $10 a month. And I I really I really wanna say this. Like, you don't have to donate, like, a thousand bucks. The typical donation that we that we got from the community for the hackathon this year was around a $100. But, you know, if you can't do that for whatever reason, because we all know life is getting expensive and more and more expensive, you know, $5 or $15 already help a long way because, you know, we're a grassroots movement and and, you know, many, many small donations also make really, really big events and really, really big help.
Okay. Well, listen.
And
I'm doing it right now. I expect everybody listening to do it. That's all I'm saying. It doesn't have to be what I put it. But but, I mean, if you...
If you're using this technology, it's been for free for a long time. And I I guess if you don't understand that... I mean, you're you're... Listen. You're a real person.
You know what I mean? Like, you have... You already have a job. You're doing this thing. I'm gonna get...
I'm gonna ask you in a second why you're doing it, but you're doing this thing in your own time and so are a lot of other people. And if you all getting together once in a while and and having the opportunity to do this moves things forward. I mean, what... Do do it again. How many CGMs will it work with now?
So it used to be you had to have Omnipod dash to run loop or to run trio, but now you can use Omnipod five.
And... You can use... Let me let me quickly try try to get all these together. So you can use Omnipod Eros, which is the the old school classic Omnipod, but you need a linking device for that. You can use Omnipod dash.
You can use Omnipod five. You can use various mini med models. You can use the Dana pump, the the Dana I and the Dana RS. You can use the Metrom pump, the Metrom Nano Patch pump, which isn't in The US, unfortunately, because it's a wonderful pump. It's it's a small patch pump.
It's smaller than the Omnipod, and it has a steel cannula. It's awesome. Mhmm. You can you can soon use the Tandem Mobi. You can use all of the Dexcom CGMs.
That means g six one, g seven one plus and Stilo. You can use the Freestyle Libre two and two plus. You can use the Freestyle Libre three and three plus. You can use the Eversense e three, the Eversense e three sixty five, the EchoCheck smart guide, and I'm probably butchering something and missing a sensor here. But, yeah, it's...
You you you can use a lot of pumps, and we are going to add... Like, right now, we're working on on directly integrating the Dex Com g seven family into Trio. So you could, like, pair and use the CGM from within Trio, and you don't have to, like, take care and maintain two apps running side by side anymore. Yeah. And we're also working on...
I was actually, when we were meeting in in New Orleans, Scott, I was wearing four CGMs at that time because we're working on, like, integrating CGMs into the app, and I was, like, testing them on myself because I'm the guinea pig here. But, like, we are... We're gonna add a bunch of new devices. And to be completely honest with you, I think since we've started Trio, we must have added, I think, six devices to the to the... Formerly, it was only Dexcom and Omnipod basically and some mini med pumps.
Yeah. My point's gonna be that you just listed a lot of different devices and each one of those took a number of people's time and attention and, like, for those of us that don't understand coding, you know, like, we're never really gonna understand what what it is you're doing. Mean, the closest I come is that my son does this kind of work and sometimes he, like, comes out in the morning, looks at me, and he's like, I have to try to connect this API to this thing and it's gonna pull from these couple of... And he can see him, but he just disappears into his room and comes out ten hours later. I'm like, how are you doing?
He goes, I I I made it work. It's good. And I'm like, oh, okay. You you know, like... And I don't know what the hell he did.
There's there's a lot of screens, a lot of keyboards, and he seems to be very tired when it's over. So, I mean, I can't imagine how many people or how much time it took to get through all those different devices and and to make them,
you know It... Like, you you would think it takes a village, but, you know, the the village is quite small. And and people peep... People, like, invest a a ton of time. Like, some of us, like, you know, almost almost have, like, a second full time job doing this, basically, but it's unpaid.
So Yeah. It's it's very altruistic.
Listen. So now I gotta tell you now that we're involved in this. I tried to make my donation, and it errored at the end. So now I gotta do it Oh, it errored. Now I gotta do it again.
But don't give up if it errors. Hold on a second. I tried to use Apple Pay. I should have known Apple Pay is notoriously terrible.
So Sometimes sometimes Apple Pay can do that to you. Yes. That is very true.
I'm gonna try one more time. I'll try a different way just to... Because now I feel like somebody's gonna listen to this and be like, did he really do it? I'm doing it. Relax.
I gotta put my email address here.
I can leave a little note for everybody. I'm definitely gonna say thank you in my note. Yeah. Hey. Well well well, while you're doing that, I can I can, you know, I can talk a bit about more about the Trio the Trio release?
QuickPick, and the Apple Watch35:03
So, actually... So many people have been using, like, the Apple Watch with with Trio, and you, you know, you can you can... We have... There's a companion app on the Apple Watch. We also have Garmin complications now.
So, you know, if you're, like, into into running or just, like, lock... Like to wear a Garmin watch, a smartwatch, you can have, like, complications on your Garmin now. It's very handy. We, actually, Mike is preparing for a marathon, and he, you know, he set up a really cool complication for it where he can see his pace and his his, like, heart rate, but also his glucose and his IOB, which is pretty handy when you're, you know, when you're running. And then another, like, really, really cool feature, and personally, I really love that feature.
I have to I have to say here. My good friend Magnus out of Norway, he implemented this feature. So the way I use Trio is I basically don't really do a lot. I I might maybe, you know, announce a meal, and I might maybe give, like, a one two unit bolus to, like, get stuff going. Mhmm.
But I don't really I don't really do that. And I... And you've met me. Right? I mean, I'm I'm I'm a gym guy.
I I work out a lot. I go running. So, like, you know, I I eat a pretty, you know, balanced diet, I would say. And so most most times, I actually, like, eat the same kind of, you know, carb amounts. And I I kinda got tired, and I guess Magnus as well, and we kinda got tired, you know, to, like, always tap the plus button and, know, tap in, like, 50 grams of carbs and then do the whole thing.
And so what he came up with is he... We call it the quick pick treatment. It'll come up in Trio if you long press the plus button in the in the home navigation. And what it'll do is it'll basically kind of learn what... How you're using Trio.
So it'll go into your data of, like, the last ninety days. And then for that time of day and that day of the week, it'll give you up to three picks. So that's why it's called QuickPick. And it'll give you up to three carp amount picks and three bolus picks, and you can just go tap, tap, tap, and you just added a 100 grams of carbs and five unit bolus, and that's it. And it's basically a three tap meal entry and a three tap pre bolus if you want to do a pre bolus because you don't really have to if you don't want to with Trio.
And it's based on your data. So if you want to if you wanna say it, Trio kind of learns like how you use the app.
Okay. That's amazing. Jeez. And and do you guys like... Do do you find yourself sitting around often and saying like, hey.
You know what would be nice? Or are you... And you work on new ideas, or are you mostly working?
All all all all the time.
Okay. Okay.
All the time. Honestly, we were we were... Like, just this week, we were on a Zoom call with, like, the community because when we do these we do these scheduled Zoom calls a couple times a month where they are never recorded, and you can just come up and and ask questions because, you know, there are no stupid questions. And people are just much more comfortable when they can, you know, just blurt out a question. And it's not recorded, and, you know, someone could make a stupid comment about it.
And so we've been doing these calls, and, obviously, we were doing a a call for, hey. This is the Trio one that I'll release, and this is the new lube release. Come up if you have any questions. Mhmm. And a guy was coming on there, and, I mean, I I don't wanna name names.
I don't wanna make anyone uncomfortable, but he mentioned that he has a visual impairment. And he was asking, like, hey. Could you maybe, you know, make the text on the the live activity, the live activity? So this is, like, the the really cool thing you have in... On your lock screen that updates live.
You might know that. Like, when you when you order, like, an Uber or a Lyft or, know, Uber Eats, you get this little thing on your lock screen, and it it kind of, you know, shows you how far out the the guy, the delivery driver is with your food. You know what I mean? The thing on the lock screen. Yeah.
We we have that for Trio. Right? And you you can you can customize it to a bit. You can have, like, a chart on there, and you can have it, you know, show you your your ILB, your glucose, the last time the algorithm ran and stuff like that. And he was like, you know, can you maybe make this a little bigger, and can we maybe make it a bit colored or something?
He Went to Bed, Woke Up to a Fix38:51
And we got off this call, and, you know, the team was still chatting a bit. And I went to bed because it was, like, 1AM in Germany. And by the time the next morning, I woke up, like, one of our US folks had done a pull request, which is, basically, he he kind of put up a new feature and was like, hey, guys. Look at this. I made fire.
And was like, hey. I I... So I coded this thing where you can now customize the font size, it's, like, super big, and, you know, it's colored. And I'm testing it right now, so it's it's not yet merged. By the time this episode comes out, it might actually be in the development branch of Trio, so it's not, like, in the main release.
But I just wanna, like, give an example. Like... So, you know, we were on a call just a couple days ago, and someone mentioned a bit of an issue for them because, you know, they live with diabetes and they unfortunately have a visual impairment. And one of the devs was on the call, and he just went, man, I'll just... Can solve this now, like, right away.
And I went to bed, and he went to work. And by the time the next morning I got up, we had a solution for this. That's something. And, I mean... And now it's in testing, but a lot of the things that you will find in these apps are basically built like this because this is patient led innovation basically.
Yeah. Well... And and you guys are having like in the moment concerns that you can... So you're you're you're almost in the same situation that I am, which is like, I just do this by myself. If I have an idea, I say to myself, I
have
an idea. I'm gonna try this and I do it. And if it works, it works. I'm like, great. If people don't like it or it doesn't strike people the way I was thinking was, it's like, that's no big deal.
I didn't hurt anybody's feelings. I only wasted my own time. But then a lot of things that I make or put out, like, actually end up helping somebody. I don't have to go to a meeting. I don't have to discuss it with people.
I don't have to, you know, oh, is Karen already working on something like this? I don't wanna step on anybody's toes. Like, none of that stuff comes up. And it's... You're just able to just pivot and and be flexible.
And and you're also in the space where you see the problems first hand so you can you can react very quickly. I imagine that's what happens when you all get together too.
Yes. Yes. Very much so. And and to to be fair, I mean, we we work super collaboratively. Like, it's it's it's a team.
Like, you know, it's it's not just, you know, I don't know. You met James. Right? You met James at 88. It's not just, like, you know, Mike, James, and Dan doing stuff and, you know, they don't talk to people.
Actually, like, we we we listen a bunch and, you know, we read along in the groups and we have we have friends in many, many other groups. And, you know, we we read along. We read more than you might think. Mhmm. And then, you know, we come up with ideas, and then we discuss them in the teams.
And and that's how, basically, feature requests come forward. And the thing with open source is, basically, anyone who has an idea can come forward and be like, hey. I have this idea. I don't know if it's great, but I just had this idea. And, you know, what do you guys think?
And then we had a bunch of things that that came came to be like this, and and and and that's that's how it goes, basically. Yeah. One one of those would be probably... So there... There's a companion app.
I don't I don't know if you're actually using it. It's called LoopFollow. You you might be using it because because John Fawcett was the guy who who, like, created it back
in the
day, and it's now being taken over because, you know, John pulled a bit out of the open source community Right. Because he's concentrating on Sugarpixel. But LoopFollow is getting major releases. Like, it's it's now in version seven dot one, actually, you know, and and, you know, John was doing one dot o just to, you know, show you how much it's evolved. Right?
Yeah. And and there are a bunch of, like, things in that app where parents, for example, just were like, hey. I I really would like to have this kind of feature to get, a certain alarm, for example, or to be able to, you know, better better interact with my child, for example. And the team just gets together and has an idea, and they solve the issue, and then we release it after a a shit ton of testing. That's the important part here.
It's not just, you know, a couple of people getting together and, you know, coding something up, and then two days later, it's released. It goes through, like, a ton of testing, a ton of automated testing, a ton of structured testing by humans. And when it's good, it's good, and then we ship it. Mhmm. Yeah.
I I use LoopFollow every day. So...
Oh, you do? I'm sorry.
Yeah. Yeah. Cool. 100 a 100%. I I...
It's the story I was telling you earlier. So I was I was sending a bolus through that.
Oh, yeah. Yeah. Yeah. See. See.
Yeah. And and so so one of one of the one of the main things, for example, that that started with the Trio team and the LoopFollow team collaborating is the way you can now do remote carve entries and remote boluses in LoopFollow with Trio and Loop and how you will now get a little push notification back whether it worked or it didn't. Like, that whole system was completely redone from the ground up because we were like, yeah. Like, you as a caregiver or as a parent, you just need to know if if stuff actually works and, you know, the the team came up with this idea. Someone in The US actually had a really cool idea and did it like a prototype, and then the prototype wasn't it wasn't scalable.
What I what I mean by that, it wasn't it wasn't so that you could just, like, build it and use it. It was, like, very complex to set up. Then, you know, the team got together, and we we we innovated, and we we tried around, and we fiddled around, and then that's what's been in for a long, long time now, for example.
That's awesome.
Also, in in the new LouFollow app, for example, because we collaborate with the team across across the teams. Basically, we're we're one big family, so we we all work on the same things. You can now, you know, pinch the chart and you can pinch Zoom the chart and you can you can tap in the chart and you get stuff in the chart. That originated in, like, work that we did for Trio one dot o. And, you know, I was talking to one of the main guys that is doing Lufolio.
He's called Jonas. His son is a type one diabetic, and, you know, that's why he's involved in the community. And we were just talking, and I I still hadn't, you know, finished my own work. And I was just... You know, we were just chatting away.
And and I I sent him a link to, like, my code, and I was like, hey. Can you take a look at this? What do you think? You're like, maybe you can use it. And, you know, and now it's in both apps, and we we we we pushed innovation, and we we made things better.
I mean, improved the, you know, the user experience when using the charts in Trio and and in loop follow. And that's just how we roll. It's it's really, really cool, and it's it's one of the main reasons that I guess I'm also involved because you can, like, you can really push the boundaries, and you can really make lives of people better. And and it's it's it's a pretty giving giving thing, honestly. Yeah.
For for all the, like, sleepless nights and all the stress and all the, you know, hardships that it's that it's kind of creating, in in one's own life, to an extent, it's it's very, very much worth it, if you, you know, get to see how you can, affect people's lives for the better.
Right. Hey. Listen. I either have donated $500 or you have my entire four zero one k. I'm not sure what exactly just happened, but good luck with whatever I sent you.
God Hey.
I mean, for for the Knightskot Foundation's sake, I kind of hope it's your 401 k. Seriously? For your retirement, I don't.
I either just donated once or twice. I have no idea. Enjoy the money. I'd I'd... You know, I hope it it flies a couple.
Thank you
very much. Yeah. Yeah. You're welcome.
Appreciate it. I'll I'll I'll I'll tell the team that you did this.
I think I'm gonna stop spending money while I'm
talking to people. I don't think I'm good at that. Hey. I mean, I can give you my address if you if you wanna go and have them, like, you know, order something. Send you over some paper towels and toilet paper or something.
Yeah. Please. Yeah. Do it.
Well, I... Listen. I mean, you've had diabetes three decades. Right?
Coming up three decades. Yes. 2027 will be my thirtieth. I mean,
if it doesn't mean a lot to you, who's it gonna mean a lot to? Right? And and I would say too, you you probably used other systems before you did this. And you you know what I mean? Like, you know the impact it has on...
Listen. I know the impact. Look. I'm... I don't think this is awkward.
I Take an Ad From Every Pump Company46:36
I think I take an ad from pretty much every pump company that exists. And I am I am more than happy to, like, tell people what's out there for them, and I hope they find something that fits them well. I also think that, generally speaking, most people are not gonna be comfortable building their own app, or being... Or or doing this. I think for the people who are gonna find it, interesting, they're gonna they're gonna go and work on it and think like, oh, this is really awesome.
And there are gonna be plenty of people who are like, yeah. It sounds nice, but not for me. When that's your world, maybe it's the mini med flex, maybe it's Control IQ, you know, and their offerings, maybe it's Omnipod, maybe it's Beta Bionics. I don't know what you're gonna like, but I want you to know about them and have your choice. I also think that every one of those algorithms works really well.
I think I I... You've been around diabetes for thirty years. My daughter's had diabetes since 2006, which is twenty years that I've been aware of diabetes. It is absolutely... Every one of these pumps is far and away, and it's sincerely better option than what my daughter had when she was diagnosed.
And people are having fantastic outcomes. I think other people also have levels of effort they wanna put into these things, and I think that each one of these offerings offers you an ability to put in the kind of effort that you wanna put in. And you Absolutely. Yeah. You should go check under every one of them and figure out which one's for you.
You know? Yep. So...
Mister Potato Head for Diabetes48:01
And and and to kinda double that on that, and and that's the beauty here. It's like, we... We're not even saying, you know, any of the commercial systems doesn't have a place. You know, every every every commercial system is is crappy. Like, don't use it.
No. Not at all. We... Like, people came up with the algorithm that we're using, and for many, many people, it's it's creating, like, much better outcomes and results, but it does also mean you have to, you know, tune the system, understand how the system works. And many people are just, you know, they aren't really comfortable maybe with using a system that isn't quote, unquote, you know, officially regulated.
Yeah. But at the end of the day, and that's... I I was talking about this earlier, what we really wanna do is we wanna we wanna be the mister potato head, basically, basically, for diabetes. We wanna be, you know, the system where you can just go, hey. This is a pump that I really like.
This is a pump I can afford. This is a pump that is, you know, available to me under my insurance scheme in my country or whatever. This is the CGM that I can use and that I really like and that also, you know, gives me good results. And, you know, I'll just use that pump and I'll just use that CGM, and I can I can fortunately pair it with the system and run it with their algorithm, and that algorithm is also really good? But that doesn't mean, like, at all that, say, you know, using...
I don't know. Using the o p five over Dexcom g seven is a bad thing or or using the the Beta Bionics eyelid because you're just someone who, like, at all, like, doesn't wanna think about diabetes, and you just wanna, like, pop something on and, you know, maybe push a button that you're eating a meal right now, which is what the outlet does. Right? Very valid. And there is no right or no wrong here.
Like, what works for you works for you.
Right.
Like, I I really I really wanna, like, double down on this because at the end of the day, diabetes is a very, very, like, personal and individual condition. And, you know, however you wanna live your life, you live your life, dude. And and I I wouldn't wanna be dictating to anyone what they have to use. The thing here is I just wanna give them give them choice, and I wanna give them options. And I wanna have them...
You know, if they use my option, I want that option to be the best option that they can use. But at the end of the day, it's just giving them true interoperability and giving them the best algorithm that we can develop.
Well, listen. In a world where I heard people throwing the word interoperability around for years and years and years, I didn't see it come. You you know? So, like, it was talked about. Yeah.
We're gonna do it, and it's gonna happen. And it's still it's still business when it, you know, when you... When it all boils down to it. And, you know Yep. I think you're...
You've come the closest so far to to offering that to people. It's all... It's fantastic. You know what I mean? Like, it really is fantastic.
I I... It's not gonna happen the other way. It just isn't. Like, you're not gonna get companies to agree on everything they need to agree on. And, you know, I hear people say sometimes, like, well, why doesn't this pump work with that CGM?
This pump company's gotta get it together. And I'll think, are you sure it's the pump company? You you know, like, maybe it's the CGM company. Or why doesn't the CGM... Maybe it's the pump company doesn't wanna work with...
You don't know what's... I don't either, by the way. But, like... But you don't know what's happening. And to think that somebody's gonna sit down and hammer through that, they're not going to.
You you know what I mean? Like, it's it's never coming. So...
It's it's it's it's also... I mean, it's it's... At at the end of the day, these these companies have to go through, like, a ton of hoops and and, you know, bringing a product to market is is a really expensive thing, and it's also just, like, you have to jump through a bunch of hoops. You have to do, like, various clinical trials. You have to go through regulatory.
And, again, I wanna... I don't wanna say that the things that we develop in the open source space isn't isn't safe by any means because at the end of the day, we are using, like, you know, fully tested, fully regulated, you know, commercially available products. We just make them available to you through, you know, going through the communication protocol, But it doesn't make them any less safe. That's that's that's the beauty of it here. But, you know, we don't have to do that is what I'm kinda trying to say here.
Like, we don't have to spend, you know, $2,000,000 on some kind of randomized clinical trial to, you know, test out the pump that we're gonna bring to market. So it it just comes with different challenges, and it it just goes to show, I think, that, like, how complex this space is. And a really good friend of mine, he said the... He said at a panel we were at... He was he was a panelist, and he was saying it's it's the fastest, slowest industry.
How so?
Fastest, Slowest Industry52:26
And... Well, because because we do move and, you know, you've you've probably also heard the whole thing of, like, you know, the cure is five years away. And, you know, I was diagnosed in 1997, and back then, it was, like, five years away. And and, I guess, in, like, in '28, it's also still gonna be five years away, the cure. But you have, like, these these major players in the market, and they're, you know, trying to fight for their share.
And no one really wants to, you know, have other people sit with them at the table and share the food. And that collectively... And I don't wanna, like, nitpick, like, single company here and be like, yeah. This is a bad company. But, collectively, that is slowing down innovation.
And then again, if you look at, like, what has happened in the last, you know, ten, fifteen years in our space, there's been, like, so much progress and so much innovation even in the commercial space. Like, if I if I look at what I had available when I went to high school and what, you know, high school does now have available, for example Oh, yeah. Just, like, you know, twenty years later, It's it's it's insane. I mean, you you were saying it yourself, like, couple minutes ago. Like, it is insane where we've gotten to.
But then again, it's, you know, it's the it's the slowest, fastest industry.
Yeah. There's a push and pull. Right? Because a a company a company is a company, and the people that work there are trying to make money and get paid and and probably make, you know, investor, you know, value and everything else. And and if it wasn't for them wanting to do that, the thing wouldn't even exist.
So you can be mad at them for wanting to make money at it, but you could also say, like, pretty reasonably speaking, if they couldn't make money at it, I bet they'd take their Legos, go home, and build something different with them. Like, they're not people who are either going to make a CGM or make a pump or not. They're gonna make a thing that they can sell or make a different thing that they can sell. These are people trying to make money by producing something. I take the perspective of, I'm super happy they're making CGMs and and insulin pumps.
Like, I'm I'm thrilled they're out there doing that. I think that you're not gonna... They're not gonna be an altruistic... You know, they're not just... They're not a...
They're not an org. You know what I mean? They are a company. They're doing their part. You're doing your part.
And I think everything helps and pushes and and assists each other. Like, there's not one pump or CGM manufacturer, or for that matter, the FDA, who is not aware of what you're doing. Like, do know what I mean? And so Yep. Yep.
It... And the whole world hasn't burned down. Know You what I mean? So it feels to me like it's an unspoken everybody's kind of okay with what's going on situation. I don't know I don't know anything for sure.
I'm telling you from the outside perspective. It's how it feels to me. So Yeah. You know, it... And and a lot of the world works like this.
It... It's it's... I know people wanna put things in, like, neat little boxes and say, oh, that's this and this is that, and it's all sanctioned or... Most of the world doesn't work that way. So I like it.
I like what everyone's doing. Everybody just keep doing what you're doing. Scott says it's good. Keep working. You you know?
Like, you guys... I mean, honestly, man, you guys made an algorithm. That thing's insane. I love it. I couldn't possibly love it more.
I talked to a person yesterday. Loved their Omnipod five. Two days ago, somebody who's... I talked to loved the twist pump. I talked to somebody who loves the new...
There's a guy on my... There's a guy in my group. If he loved the mini med flex anymore, I think his head would explode. Like, do know I mean? Like, it's...
He's so happy with it. And Yeah. Yeah.
And it's good and it's good that he's happy with. And Yeah. You know, we, like, so so, like, our... Again, we're standing on the shoulders of giants here. Like like, our team owes it to to Scott and Dana.
So Dana Lewis and Scott Librand who who came over the algorithm that we are now using in in Trio, and it is also being used in Android APS. We're standing on their shoulders. Right? We... We're also standing on on on shoulders of other people that did other cool things in our community, and it's it's the same thing of, like, you know, we...
I don't know. I mean, I don't wanna get, like, too philosophical here, but, like, if if if it hadn't been for banting, you know, and and, you know, basically making insulin patent free, we would still be dying like flies. Right? And so it's it's it's clear where we've what we've, you know, where we've gotten to. And I I I just hope that everyone just, like, you know, gets access to the things that they feel like they can use and make their, you know, their life suck a little less because they have diabetes.
Like, know, make make diabetes suck a little less.
In the sixty third and sixty fourth episode of this podcast, keeping in mind that your episode will probably be, like, episode 2,020 or something like that.
Okay? Mhmm.
I talked to Dana and then Scott back to back. And I haven't heard them in ten years, those conversations. But I bet you it's just me going, I don't understand. What'd you do? Yeah.
Right. Right. Yeah. And then it's in a Tic Tac box. What's in a Tic Tac box?
Your Raspberry Pi's in a Tic Tac box. I don't know what the hell you're saying. And, like... And that and that was literally, like... I mean, honestly, it's, twelve years ago.
And, you know, and and to think that we're sitting here now, I I mean, I couldn't be happier with how Arden... Arden uses Omnipod with Trio and Dexcom g seven fifteen day. It's fantastic. And I'll... And let me add this to it.
We're not talking about it. Little GLP really writes the whole ship. Okay. So... Yeah.
Yeah. An algorithm, a CGM, a pump that can break up that insulin, insulin, and and a a GLP, you could you could sleepwalk through a six a one c. And and Yeah. That's the that's the
For many people listening.
For many... Is it every... It's not everybody. I'm not saying... I'm saying, like, you know, for a mass of people.
This is this is fantastic. I still... I just interviewed a woman three hours ago who was just diagnosed as an adult, but started talking about how her sister was diagnosed some twenty, thirty years ago as a baby. Her sister is still struggling because she came up with die... In diabetes in a time where somebody told her this is a sliding scale.
Here's how you do it, and she just never progressed. Like, so Mhmm. Where you put people is sometimes where they stay. And imagine that today where we're putting them is in all of these algorithms that exist on all these different pumps with all these different CGMs. And and that...
Are you telling me that's gonna be that person's baseline twenty, thirty years from now? A six, a six and a half a one c eating how they like, living their life fairly unencumbered compared to the past. You talked about what you had going to school. My daughter was diagnosed, given a a... Literally a bag.
We just had a box with bags inside of it. The bags had 10 needles in them. They gave us Yep. They gave us basal and and bolus insulin and a meter that that literally felt like it fell out of a bubble gum machine. And and that was the entirety of how we manage Arden's diabetes.
And if Hey. And, I mean I
Twenty Years of MDI and Finger Pricks59:13
mean I mean, it's it's it's... I was the same, basically. Right? And so we we... So, I mean, this is the second episode.
I've I've been on your podcast before. We kind of... We we kind we kind of touched on that in that other episode. But just to quickly quickly touch on that again maybe, I was doing MDI and and finger pricks for twenty years. And so if you if you think about this, like, I was still not wearing a CGM in 2017 because, you know, my endo was this old guy.
He's a type one himself, and, you know, it was kind of like, you know, don't don't turn the running... Like, don't don't turn the ship. It's it's going it's going great. Like, you know? Yeah.
Set the sail and go ahead. And I was having good a one c's, but holy holy freak I was doing so much work. I mean, if you wanna if you wanna get to, like, a, like, a five nine a one c and eat a normal diet, do you know how much you have to, like, micromanage on MDI and finger pricks? Oh. Holy.
Oh, I know how much because I did it, like, for a bit...
For a
kid, and it was killing. Right? Like, literally killing me. So Yeah. Yeah.
Yeah. Yeah.
And and and so and so I was... I I I got onto a pump because of of some other medical condition. I was, you know, I'm... We we made the... We made the switch to a pump.
And at that time, I was I was also already working as a software engineer, and I had been, you know, following along the open source AID community, but I didn't have the devices. So I was just kind of, you know, sitting sitting back in awe. And, dude, I went I went to my Endos clinic, and this nice lady from Insulet came up, and she was like, so this is a PDM, and this is how you use this. And, you know, she gave me the pod, and I filled up the pod, I put the pod on my arm. And then I I, you know, I was I was nodding and smiling.
I was nodding and smiling, and then I signed the piece of paper that she really taught me how to use it. And then I went home. I pulled off the pod. I filled up a fresh pod, and then at that point, I was running loop because back then, there was only loop, basically. Right?
And, you know, fast forward a couple months later, I started... I joined this team and started developing, like, our own app. And, again, fast forward fast forward a couple of years, I'm doing what I'm doing now, and I'm sitting on a podcast with this guy who's doing podcasts that's, like, millions of people listen to because I'm doing this app. Right? So...
And not only that, but I'm making the podcast the same way you're helping with Loop. Like, I just one day was like, oh, I live in this. I think that I have a talent or a skill that might help with it. Let me see what I can do to help. That's really all I was thinking to doing, and you're doing the same thing.
It's... And and there's a lot of people doing stuff like that. A lot of people who nobody knows, and that... That's always my point. So, you know, people know me and they know this podcast, and that's wonderful, but there's a lot of people doing a lot of stuff you have no idea about.
Also, you know, I also know a ton of people at every one of these companies. They're also working really hard and lovely people. And many times, a lot of them have diabetes. And there's just people in corners of this world, nooks and crannies trying to help you with your type one and trying to get us as fast as possible from here's a bag of needles, good luck, to where we are now. Also, I I guess it kinda takes me into the next part of my conversation.
Let me just tell people before I jump here that you were initially on episode 1,243. It's called DIY insulin algorithm a I a p s. So if you wanna hear more conversation, go check there. But it it... Taken that we got...
We've gotten as far as we've gotten since my daughter was diagnosed in 2006 to where we are now. Just from your perspective, as somebody who's working on something specific, where do you think this is leading to? Like, I mean, how much is new technology going to help you? I mean, is there a ceiling on these algorithms? Or or...
You know what I mean? Like, are we gonna get to the point where it's just like when we buy an iPhone, we're like, the camera's better? Like, or are... Is it gonna is it is it gonna hit a ceiling, or is it going to... Do you think it can continue to prove?
Is There a Ceiling1:03:04
So speaking of what is available commercially, there's... There is a lot of ceiling left to to... Until we, like, hit the ceiling. There's a lot of air. Speaking of what we're working on, there's also still a lot of air.
I think right now, if if we take, like, three steps back here. Right? So there's a bunch of research going on. You know, there's, like, antibody medications now going on that, you know, can postpone the the the breakout of, you know, the Sure. Beta cells being killed and stuff like that.
And I guess that is that is, I guess, the actual cure to an extent. I hope... I still hope that at some point, we will get to that actual cure. But what I personally like to call these systems is a functional cure. Right?
So right now, this is, I guess, the best thing that we can get, and we we are still trying to improve them to to having to do, you know, less less inputs. So, you know, just eat, and you don't have to worry about boluses, and you don't have to worry about putting in insulin. You don't... Ideally, you don't even have to worry about, like, you know, what you're eating exactly. So that is, I guess, what we're working on.
And then if you wanna get a bit more detailed, you know, right now, we are working a lot on stuff like there is something that, you know, auto adjust settings, and there is something that, you know, kind of maybe like a questionnaire, and you you you answer, like, five questions, and it sets up the system, and it it sets up the settings for you. Because one of the main pain points for many people that use an insulin pump is setting up the system and finding the right settings on that system for them. Right? I I think the the the ceiling that we'll hit at some point right now, other than, you know, improving the algorithm and making it better and better, is a, that there is currently nothing like a smart insulin. So, like, insulin dose is insulin dose.
You can't really take insulin out at this point. And the other limitation is honestly just the accuracy and the quality of the of the, like, the input that we need, and that is glucose readings. So how good are CGMs working? How fast are they working? It's it's still, you know, based on interstitial fluid.
So it's not actual blood glucose. It's it's, you know, it's a chemical reaction based on the fluids in your cells, basically. And because of the chemical reaction, they can they can calculate what glucose you will have, but it is still delayed. Right? So I guess those are the the limits right now.
But on on on... Like, we... There's still so much to be done and so much to be improved, and I I don't think we're we're we're there yet, honestly. And also, taking another step back here, you and I, we are very fortunate enough that we were born in, you know, the Western world in in, you know, in very educated and rich countries. Right?
Like, you were born in The US. You live in The US. Your kid was born in The US. I'm I'm I'm here in Germany, Central Europe. We have, you know, very, very good medical care.
We have access to all these systems. But there are, like, low and middle income countries in the world. Right? And I think one of the one of the things that also the open source community is working towards is really making these systems available to as many people as possible for... You know, to to to just also have some kid that is living, you know, in in rural India somewhere or, like, I don't know, in in somewhere in in in the in The Middle East be able to use this kind of stuff.
Yeah. So, you know, I don't wanna I don't wanna go into politics at all, but, you know, it's it's it's a crazy world we live in, and I think there's a lot of things to be done for, like, people with diabetes in the world Yeah. Generally.
Yeah.
Right? And and then when it comes to these systems that we're using, I think there is still a lot to be done. And I don't wanna, like, sour the mood here at all, but it's like we we are still very, fortunate that we can we can use these systems. And, like, for me and for Arden, it's something like, you know what? I'll just have lunch.
I'll... I don't give a And for someone else, it's, oh my god. I don't even have insulin for the next week. Oh, yeah. And I think I think this is like the...
It's it's it's like a really double edged edged sword. I'm sorry. But I think we'll we'll improve as much as we can, and we'll also very much push to try and make this available to as many people as we can. But I think the... I guess this...
I guess the ceiling to an extent is trying to make life with diabetes better for everyone that has diabetes on this planet and not just, you know, the people that are fortunate enough to have access to these systems. Yeah. Is there... Sorry. I don't...
I didn't wanna sour the mood here.
You know what? You know what? Don't you sour the mood. It's it's absolutely true. I...
Listen. I've interviewed a ton of people from The Middle East who have, like, financially the, you know, the the ability, but don't have any access at all. A lot of families, Saudi Arabia, places like that, they get on a plane and fly here and buy their their devices for their kids and then fly home again. Like... And, listen.
It's amazing they can do that, but the point is is that they live in a place where they have money and they still can't get to it. Imagine places that... Where you don't have money and you can't get to it or you don't have money and you you could get to it if you had money. I mean, most people are not enjoying this kind of this kind of technology or outcomes, even access to information about something as simple as like, hey. The fat in your food makes your blood sugar go up in a couple hours.
Like, that's... Yeah. That's just not a thing people talk about. It's... I hardly find people in America who were taught this stuff.
The podcast is, generally speaking, popular because I talk about prebolising and covering, like, macronutrients that impact your blood sugar. Most people don't talk about that. And then they have, like, wildly better outcomes because they found out about that, and then they think I'm, like, special. I'm like, I just... I told you some pretty basic stuff.
You know? But you have to you have to have somebody who's gotta be willing to have that conversation and tell you about it and go over and over again until you're like like, oh, I'll try it. Like, leave me alone. You know what I mean? Like, until you get kinda, like, pushed into giving it a shot one day.
And if you're... I don't know if you're digging in a mine for coal bought, though, there's definitely nobody talking to you through a podcast about the fat in your meal if you're lucky enough to have a meal. Alright. I'm with you, man. You know?
Yeah. Yeah. I I... But but what you do specifically, if those people can't get to the technology, the thing you're not... The thing you're doing is not going to be able to impact them.
Is that That
is correct.
Is that hard for you? Like, I mean, do you feel like you're helping and you could be helping so many more people, but there's things in your way that are out of your control?
I mean, obviously, there are things out of my control. Right? Or, like, out of our control. But, it's it's it's a growing thing. So, like, even in in low and middle income countries, more and more people are using open source AID because it's it's...
Again, it comes it comes to the whole interoperability bit. Right? If if they can get their hands on some kind of pump and some kind of compatible sensor and then they have a phone, it it... They have a system that works for them. Right?
And so this is this is, like, really, really awesome to that regard. I think the the limitations that that we kind of struggle with is just, to a point, getting the word out there about these systems. Because I think at the end of the day, if you have access to all of the, you know, all the little components that you need for them, they can they can, like, just massively improve your life. And and then just also, like, just advocating and educating about these systems. Right?
So open source AID has been part of the ADA standards of care for, like, a couple years now. A trio is being mentioned by name for the last three editions. Right? The the... There was, like, a massive consensus statement published in The Lancet, which is, like, a really, really, you know, known and famous publication in in the medical space where, you know, dozens of endocrinologists and leading researchers actually came out and were like, hey.
Open source AID is really something you guys should support and clinicians and, you know, practitioners should help and support their patients if they wanna use this. And so a lot of the things that we feel limitations right now is honestly just, you know, doctor Jane Doe and doctor John Doe in in the middle of wherever that don't really keep up and, you know, then their patient comes along and tells them about the system. And they'll just, like, write off, tell them, nope. You can't use this. Like, I won't support you.
And that that is the, like, kind of limitations that really frustrate me because, you know, they could very much use these systems because their doctor just has to write a prescription and be a little supportive about it and maybe learn a thing or two. And those are the kind of things that frustrate me. And I I don't really can change them. Right? I can't really change the world in in in in a...
In the, you know, blink of an eye here. But those things are frustrating, and that's why we do advocacy. That's why I travel to conferences, and that's why I'm, like, involved in research and and, yeah, general advocacy for diabetes as well. But I really hope to be, you know, improving that aspect. And then another aspect, I I shouldn't forget this actually.
Female Bodies Are a Research Gap1:12:20
I'm I'm kind of embarrassed that I'm just thinking of this now. It kind of shows that I'm a man. A big portion of of, you know, diabetes research and medical research in general, but, like, especially diabetes research is also... That is overlooked is is, you know, is just the whole, you know, female side of things. Right?
The female body is just this super duper complex thing, and we don't really understand a lot of things of it. And especially when it comes to diabetes, you know, how does how does the female cycle influence insulin sensitivity and how algorithms work? You know? How do various things in in the female body and when it comes to pregnancy and stuff like that? How do they how do they influence diabetes?
How do algorithms need to adjust for them? Do the systems themselves need to adjust for them? Just, you know, how, you know, people use them. And so I would... I I really, like, love that we're doing work in this in this realm now.
There are a couple researchers that are putting down work. Actually, there's the team out of Charity Berlin and the Hauser Platte Institute here in Germany who's working on this. There... There's a team at Typool also who's working on this, and they're collecting a bunch of data and doing research on this. But I'm I'm going from, like, one to the other here, but it's it's...
There are, I think, some systemic issues that we just can't change or or, like, systemic limitations, I guess, that we just can't change in an instant and that we're, like, advocating for. And then there's just, like, knowledge gaps, and I really, really hope that we can close these knowledge gaps and do a lot of work and research to close these knowledge gaps. And one of those big big gaps is, as I was saying, is the whole the whole female body and
its Yeah.
Its connection in relation to, like, what diabetes does to it and how insulin works and how algorithms need to work for it and stuff like that.
Yeah. I happen to love to adult women, my daughter and my wife, and to watch them try to get through their life, especially with autoimmune stuff. And it's fascinating and really disturbing how frequently a woman will get to an answer to doctor's office and the answer is we don't know. And Yeah. You know, and good luck.
So... Or you... Maybe you're depressed. And, you know, somebody... I've heard people say, yeah.
Well, I mean, if you were in pain every day for three years, you might be depressed too. And and just like how quickly they there's no answers and then no studies. I... I've been watching recently the endometriosis, like, world. And apparently, now suddenly there's a lot of study about endometriosis and everyone will joke, well, yeah, they found it in a guy.
So now we're now we're working on it. And even that, like... So we don't know for sure, but we're not... We're we're getting sure and sure that Arden might have endometriosis. Right?
And it's a a painful issue. And so I've been digging into a kind of behind the scenes and you find this person, you know, is working on this this peptide that they think is, you know, they kind of basically talks the cell into just killing itself and... Great thing. And six months ago, they make this announcement. You can't find, like, another thing about it all of a sudden.
Like, it's just... I I mean, I've reached out to the company. I haven't heard back. I've reached out to the doctor. I haven't heard back.
I'm like, hey. Is this a study? Is it happening? I'm interested to interview you about it. If so, I can't find anything online about it.
Now maybe they're just not talking right now, but there's not a lot of people moving and that's just on one thing. You know? And and some of these... Some of the pumps are like, hey. You know, we got...
You know, I I think Tandem's got an FDA approval for pregnancy now. That's the only pump that has, a pregnancy approval, right, at this point?
The Ipsa pump and the the chem APS FX algorithm that is unfortunately not available in the
Yeah.
Roman Roman Kavoka and his team actually, like, went out, and and one of the main things was, like, he was saying, I wanna I wanna develop an algorithm and a system that works for, like, women that wanna be mothers and, you know, have children and then also ideally for children because that will essentially cover everyone and not just, you know Yeah. White men.
That's people. Right? It... But, I mean, my point is is that... And I don't know for sure, but Tandem had to make a decision at some point.
Like, we're gonna spend money to do a trial to try to get this distinction. And because that's how this... I mean, you made the point earlier. Right? Like, it's expensive to get something through the FDA.
And and why do it if a doctor's just gonna give you a pump and let you use
it off label? Like, you know what
I mean? Like, so this stuff just never gets dug into. And I... It... You know, it's because it's...
Life's Not a Pony Farm1:16:48
I I guess either the juice isn't worth the squeeze or, you know, something to that effect, but it's gotta be something similar. Can I tell you something funny? I took the Go ahead. I took the the transcript from your first episode because the way I make my my episode pages now is different. And I I thought, oh, let me refresh his in case people find their way back to it.
But part of the... I I... This will be boring maybe. Actually, probably not to you, but to most people. I take the transcript of a podcast now, and I give it to my friend Claude.
And through long conversations that Claude and I have had, he turns it into a web page for me. Not only does it embed an audio player, but it gives you, like, moments where you can jump to where it jumps to the audio. It gives you the the transcript if you wanna read through it. It gives you links back to players. It actually...
You know, the the advertisers get put in there. There's takeaways. It's like a really, like, lovely little thing. Right? One of the things that that that prompt does for me is it helps me look through the transcript for possible titles for the episode.
Okay. Yeah. And so the title it came back with from yours, from episode twelve forty three, which initially, I think came out, like, the end of June or early July twenty twenty four. And it's funny because I have no context for this, but it wants to call it life's not a pony farm. Yeah.
Now I'm dying to know how that happened.
We we we... So we we were talking about, like, idioms, and and, like, I... That is the literal translation of of a German idiom.
Oh, okay.
And it would like... The literal translation for it is like life's not a pony farm. Yeah. That's what it came up with. Anyway, I just...
I... It popped up in front
of me because I did it a while ago. I have... Like, I just dropped the transcript in while we were talking. Thought, oh, I wanna remember to do this when I get off the recording. And and then it's just...
I'm looking at I'm like, what does that mean?
So Yeah. Yeah. Yeah. Yeah. I...
I'm I'm trying to think of the English version of the idiom. I I think it would be something like, you know, life's not a walk in the park or something like that.
Yeah. And I
think that's the... That's the translate. Like, that's
Right. That's interesting. Yeah. But... So...
But I I I was... I guess I never really I never really asked my question before you got melancholy a couple of minutes ago. But, will will will you... But this brings it back up again. Will you get any help from these large language models?
Will this make your work easier, or is it... I mean, in any way valuable to you, or is this this something you need to sit down and hand code and and that's just how it has to be?
No. No. It's it's... So so you... We...
So we have to make a very, very important, like, distinguish... Like, we would have to distinguish here, like... Because there's... Okay. How how do I put this?
So if if if you, as Scott, would just ask, you know, Claude, hey. Implement an automated insulin delivery algorithm for me. And then, you know, it it gives you a a bunch of code, and it tells you it's so great because it did this and that. You know? And then you just go ahead and use it.
And at the end of the day, you might really hurt yourself. Right? That is not how we can, you know, use an LM. But, yes, we very... Because, like, you know, you have to have contextual knowledge.
You have to have knowledge of the systems. You need to be very much the human in the loop and and verify everything that it gets back to you, especially when it comes to, you know, a drug with such a narrow, you know, therapeutic indication because, you know, insulin, a couple units more or less, and you can literally die. Right? So you can't just throw something at a clot and be like, hey. Do this for me.
But I will say, yes. It has helped tremendously, of course. Right? Anything that doesn't touch the algorithm and anything that doesn't touch instant dosing. So for example, something like, you know, stuff in the user interface, we can ask Claude or or another LLM for some help on that.
It's it's incredibly helpful for stuff like analyzing logs when someone, you know, sends an issue report. And the the logs that our systems create, you know, it's it's a ton of text, you know, because all the different components under the hood are basically, you know, sending messages that are get... That gets stored. And so analyzing those logs and those issue reports, it can help with that, for example. Yes.
We're also using it to some extent to to review new features and to test new features and also to some extent, you know, to, like, write tests. And then, of course, it can also help in certain people that really know what they're doing in trying to reverse engineer certain components. But just, you know, cluelessly going ahead, opening a call chat window, and just asking them, you know, to create the next best insulin dosing algorithm or whatever.
Probably not the way to go. Yeah. Do you think that's how most people think of of it that you just ask the magic box a question and it just gives you the answer back?
We have... Like, it... It's actually... I mean, good that you're touching on it. It's it's actually a bit of an issue.
Like, we have people, posting, like, pull requests or, like, offering up features, that are so clearly vibe coded where they cannot explain themselves what is actually happening and, like, you know, what is happening in the code. And a lot of them will do stuff that does touch on instant dosing, for example.
Oh, jeez.
And then they, you know, they they put out these features and then... Or they... Also, another thing, a lot of a lot of people are asking, you know, Claude to make adjustments to their insulin settings, for example. And it's it's really not a good thing because an LLM cannot cannot do this kind of work safely unless you, you know, you have a bunch of God Rates in them. And even then, an LLM is not really the the good thing to to use.
I'm actually involved with good fellow friend out of London. His name is Tim Street. He also has a blog, Diabetech. You can you can give it to Google. He does a lot of work in in the AI realm, and him and I have have also collaborated on some of this.
A Paella GPT Called 423 Carbs1:22:55
And we did a fun little experiment where... Or he did a fun little experiment, I should say, where he sent the same, like, 12 or 13 food photos. Right? So they're just, like, snaps of, you know, a plate of something or, like, a pan of paella and stuff like that. And we send it to the various LLL models, and he send it to the LMs 27,000 times.
So every image was sent to the LMs about 500 to 600 times, and then we compared... Like, he compared the results. And we had crazy stuff come out. I mean, I can I can send you the photo of the paella, Scott, after we're done here because I actually? Like, that was my paella in Barcelona at ATDD.
Mhmm. And I think I bowled this for 60 grams of carbs. And GPT was was of the opinion that it... I should have bowled this for 423 grams of carbs. That be done that Yeah.
We wouldn't be having this conversation right now.
Oh my god. No. And,
yeah, I was gonna
say there's no way it's doing that. Yeah.
But Yeah. And and and and and and the... I I don't wanna downtalk LMs by by by all means, but it's it's really bit of an issue because the thing is an LLM will make it sounds, like, super confident. It it is built to regurgitate text in a way to you that makes you trust in that answer and in that response. Mhmm.
And what you cannot see is, for example, is intermodal variance. So if you sent the same kind of food photo 500 times to an algorithm... Sorry, to a... To an LLM, it it is not deterministic in its in its... In how it interprets it.
So you will get essentially 500 different responses. And what you cannot see is is the response pretty much spot on, or is it, like, really, really, like, on the lower scale, or is this really, really on the bigger... Like, on the higher end of it? And then and then it can happen, like, it telling me, like, that a paella would be 400 grams of carbs. And and...
But you don't see that, and you just get a response where it tells you, yeah, dude. Like, oh, this is this is a really, really nice looking paella, man. And you even got gumbos on there. Wow. Cool.
Like, oh, you must be in Spain right now. Really cool. Enjoy that. And by the way, bolus, this and that. Right?
It's it's it's the typical message that you've that you've read from, like, Claude or GPT Right. Or Gemini. And then and then the thing is, right, I mean, I I... I've I've had this stuff for for long enough, and I've bolus for for, you know, enough meals that I kinda know this isn't 400 grams of carbs. But if you are someone that is, like, newly diagnosed
How would you know?
You... And, you know, suddenly you get this, and then and then what do you do with it? And so, like, my recommendation to this is be the human in the loop and, you know, get a second opinion. So is is this, like, super comfortable? No.
But, like, would you maybe, like, submit it again, or would you use another LLM and compare it with? And, you know, these are some of the things that you could do. But just like blindly trusting an LLM for, like... In our context, for for anything on the AID system and just code isn't gonna help you out as much. And on the other hand, even in diabetes, just like blindly trusting the system is is a bit of an issue, I guess.
Yeah. I wouldn't have imagined you would have. I just... I wanted to ask the question so you could answer it like that. I I find it...
Look. I have no technical skill, but I... There's places I've used it where I'm like, this is really valuable for this. And... But you also have to be able to look and go, but there's no way I'd let it do this.
You you know, like, it's... I mean, I don't know. Like, figuring out, like, a carb ratio is a simple math problem. If somebody built themselves a little tool that, like, they put in their carb ratio and a couple things and it gave them back an answer, I'd be like, alright. As long as the tool's right and the math is right, then it'll probably work over and over again.
That's not the same as asking it an obtuse question about a a picture or, you know, I'm having Yep. You know, I'm having a grilled cheese sandwich today with a handful of french fries. What do you think? Like, that's not... Yep.
Do... Like, don't do that for God's sakes.
Funny that funny that you mentioned a grilled cheese sandwich. So so one of the things that that Tim always does is is it it sends... He sends, like, an image of a of a of a of a cheese toast, basically. It's it's a cheese sandwich how you would have it in The UK because he's from The UK, And not a single has been able to crack it. And he's also testing a bunch of, like, paid fork apps and, you know, and and also free, like, carb estimation apps that all use AI and, you know, are are, like, advertised to be this great thing.
And not a single app was able in all his tests and all his publications that he's done on the topic to correctly guess the 40 something grams of carbs on his, like, cheese sandwich. No. It... And and so he's he's been calling it the end of AI.
Yeah. And also it... Jokingly. Jokingly. Yeah.
It doesn't see the way you think of seeing. It doesn't... Exactly. It it also can't hear. Like, there was this...
I... There's this moment where I was like, this is interesting. And I, like, I put an m p three, and I was like, what does this sound like to you? And it was like, I can't, like, I can't hear it. And it was it was like, I can interpret the data from the m p three and give you...
And it it was interesting. It gave me, like, a a pretty interesting idea of what kind of music it was, but it can't hear it. It also can't see pictures. It's not the way it it interprets stuff, and it and it... And for some things, I mean, like, here, I'll give you a simple example.
Right? Jenny and I, Jenny Smith, we do these episodes. We do, these bolus four episodes where we pick up a a food and, basically, Jenny, like, off the top of her head just kinda, like, guesses at the carbs and how she would bolus for it and stuff. While she's doing that, I'm using, like, an estimator that I built that I'm, like, I'm trying to get the same thing from. Right?
And we kinda compare back and forth to see how close it gets. It's a nice conversation. People like it. I I... They're incredibly popular.
And so very recently... And for people who don't track this stuff, it'd be boring for you. But when when Claude went to... Every time Claude goes up a model, I go back and I I have it relook at any code that I've made for my website. Website because I just think, like, clean it up, make it faster, look for mistakes, that kind of thing.
And it usually does find stuff. But every once in a while, you'll notice it gets better at something out of nowhere, and you're like, oh. So I I took a a a transcript from one of those conversations, and I said, can we turn this into, a visual... Like, a web page that you could listen to the conversation, but it would kinda jump around from visual stuff and maybe... And then as I was doing it, I was like, oh my god.
Remember in the chef TV show? This is gonna be, like, very random for you. But, like, the the movie director John Favreau made a movie called chef.
I I know the movie chef.
Oh, and then
It's the grill... It's it's it's the grilled Cubano of heaven.
Exactly. So so... And then and then later he makes a Netflix show. It's a cooking show. And they they used to do this thing where, like, the the food truck would...
Like, was an animated food truck. It would bounce down the road. When it hit a big bump, it would explode, and the food item would come out and it would explode into all of its ingredients. And that popped into my head. I was like, oh, we should, like, have an image of the food we're talking about and then let it pop apart into its ingredients if we talked about it.
Only if the conversation lends to the image should there be an image for it. And, you know, I messed around with it for a while, and now I have a whole web page of them. And not only not only you click on it and you can you can ask it to follow the conversation, it'll just move throughout the web page, show you different pictures, images, pull quotes from the conversation so that it's a little more interactive where you can bop around it. And, like, I think in one of them, we must have gone through, like, five or six different donuts. You could just click on the donut.
It'll jump right to that part of the conversation. You'll hear it and it's over. It's really kind of amazing. And had I tried to do it five months ago, it wouldn't have worked. And and I...
And I'm wondering what happens in five more months or six months. I'm I'm wondering, like, when do we get to the part where you say, hey. This is my trio. You know, this is how this part does this. Do you see something that it could be doing better?
Or do you think that's... I mean, when you say there's more error, do you think that's how we get to the error? Do we just get it through trial and error and human beings using it?
Machine Learning Isn't a Chatbot1:31:03
Well, so if if if you're if you're talking about just, like, algorithmic improvement, I don't think that's just technically an LLM is the right tool here. And and and that's... I guess that's that's a bit of a... I I don't wanna get too technical, but we we have AI. Right?
And AI is actually, like, a bit of a spectrum. So, like, you have machine learning, and then you have GenAI, which, like, an LLM would be. And then you have various various classification models that that, you know, also fall into that spectrum. So I very much think that we will be using more and more AI, but the the AI that we will be using is is, like, machine learning stuff. So, you know, getting a ton of data and then trying to, I guess, predict patterns out, like...
Or find patterns in that data and then use these patterns to adjust how the system functions. Right? That is not something like Cloud or like an LLM or or chatbot for that matter. You can... What what you can use an LLM for, and I think that's...
That we will probably be seeing that more and more in in commercialized apps and also maybe in Tria at some point. I don't know. Would be something like having a little chatbot be part... Sorry. Having a little chat be part of, like, the the user interface, and then you can ask it a question.
It it explains to you a complex concept. Like you were saying. Right? So, like, it it... Right now, they can listen to they can listen to something, and then it kind of, like, moves them around and how did...
You know? And then maybe you have a little chat window, and you could even, like, ask a little question, and it, you know, still moves to, like, another image and shows you another, like, set of information. Yeah. I can see something like that maybe happen... Like, happening in Trio where, you know, there are certain buttons and then, you know, a a little view opens, it's it's called the loop status view.
You might have seen it, like, when you when you tap on the the green loop circle, and then you see everything that is going on under the hood. Right? And a bunch of numbers and a bunch of weird metrics and a bunch of weird abbreviations. And something that I can see, for example, in LM being used for is we make all of that gibberish that doesn't really mean something to you into, like, a single human readable sentence. And then it'll just tell you, Scott, you're really high right now, but you also have a lot of, like, like, really high insulin.
Maybe your pump cannula is bent. Have you checked it? Mhmm. Right? Something like that.
And it doesn't mean that we wouldn't be maybe using, like, the bed... There isn't already work happening actually where, like, machine learning is being being utilized in in making algorithms better. Same guy I was mentioning earlier, Tim Street, he has been involved in the open source community for a decade plus. He he builds a little branch or, like, a fork of Android APS called Boost, and he's experimenting with a ton of things in Boost right now having to do, you know, with with identifying sleep patterns, identifying movement patterns, identifying, you know, heart rate stuff, and how he can utilize that for adjusted dosing. And there's a bunch of things happening already, right, where we use machine learning, for example, and try try trial and error around with it in in a safe space.
So with, you know, simulation in vitro, not necessarily on the human body to try and find a better algorithm. I'm pretty sure we will we will be doing that more and more and more. Yeah. But I think just the whole concept of of having, like... Everyone thinks, like, Claude is the next best thing since sliced bread.
But, yes, it it can help you a ton, but there are just, like, limitations to those systems. And we should just be, like, applying the systems in where where you can probably apply them.
No. I appreciate you talking about it. I just... I really think that most people hear it and think that you just say to it, like, make a better insulin pump for me, And and that just magically happens. But I I I I wanna drill down for a second on your idea about the machine learning because, I mean, I'll open up and tell...
I've said this before, but when... As an example, like, have Dexcom on fairly... Probably once a year. Right? And they're like, oh, it's getting better.
It's doing this. And I used to think in my heart, like, oh, they must have fixed the wire or made the chemical they put that powder. Like... And and I realized one day, I think it was Jake Lee just said to me, he's like, no. We we just have a lot of data.
And then we're looking... And he's like, when we tell you it's getting better, he's like, we're turning knobs. Like, we're, like, making... It's the it's the algorithm that's getting better. It's not the it's not the the physical stuff.
And I was like, oh, and it just hit me out of nowhere. I was like, oh, that's where the ceiling is left. Like, they've done all they can do with... I mean, I'm sure there's improvements to make with the physical device.
Yeah. I mean, they they can make the thing smaller. They can make the transmitter better so it stops losing connection to the pump and stuff like that. They can do that. Right.
But... You know? But the wick that's in there measure... You're you're soaking up your interstitial fluid so that it can be measured, like, that's pretty much works pretty well already. And when he's talking about making an improvement, he's making an improvement to the...
You know, to his... For people, like, listening, like, for the software they use and how it interprets the data and comes back with, you know, what what the... And turns that data into something in the physical world. Right? And and I think you're saying the same thing is that, if you have enough data of women who are going through their periods or sleep cycles or that kind of stuff, like, you can use you can use a machine model to, like, look at that over and over again and try to just harmonistically figure out what it means and how you can...
How it can repeat in ways that that you can trust so that you can try to impact it. Is that really Yep. That's right. Yep. Okay.
To to a point, I mean... And and and to to to, I guess, put on my my citizen researcher hat here, in the first step, collect all of that data and, you know, find how this influences, say, insulin dosing and how insulin works in the body, for example, how it affects the concept that we have coined insulin sensitivity because, you know, that's also just a thinking model, right, that we kind of made up so that we can... We we have a way of describing and working with something that is extremely complex and happens, you know, in an instant in the human body. Right? Because your your body and your pancreas doesn't think an insulin sensitivity, you know, rates and and what you just ate, it just starts secreting a hormone.
Right? But we made up this this model of, insulin sensitivity, for example, and we're working with it. And so we have this and, I guess, you know, every woman you talk to, they will tell you, like, going through their cycle depending on which phase they're in in their cycle, they will have different insulin needs. And, you know, that is, I guess, somewhat, you know, part of the of the published published knowledge out there. But how exactly does that happen?
Can you can you predict it? Can you find it in patterns? Can you can you, based off of a glucose pattern, find out which actual phase they're in in their cycle? Can you use that to, you know, adjust the dosing? Can you use that to
Mhmm.
To to adjust other things in in in the algorithm? That is the stuff where, like, research is more and more steering towards, and that is... Mean, like, the second step, I guess, that is where the algorithms will will steer towards, and and they will utilize that stuff, and then they'll get better and better and better. And and then again, to, you know, kind of reiterate on the whole concept of we are pushing the envelope here, we will be we will be at the forefront of that, and we will be pushing that that that envelope because we can do it much quicker and, you know, from from, like, a patient leading perspective and approach versus what all the commercial companies and the commercial entities are doing. Yep.
And the great thing is that we've... We were now working, like, more and more with researchers and being kind of seen as as a serious, you know, entity to be... Like, sorry, as an entity to be taken seriously and to, like, work with versus, you know... I think it must have been actually a former CEO of Taxcom who calls the the people that created Nightscout the rogue cowboys. Right?
We're not rogue cowboys. We're actually, like, pretty pretty serious people, and we're pretty serious about the things that we do here. And we we we wanna innovate, and we wanna push the... Push stuff. And there are researchers and clinicians out there and also people in in other areas out there that start realizing that, and so that's what we'll do.
Did they ever come and try to hire you guys, like, different companies? Did did they
Well, not me personally. I'm... Might have gotten a couple approaches, but it's it's not like we don't talk to the companies. Let's just say it like that. Like, we we know people at companies and they talk to us.
And as I was saying earlier, a lot of people at these companies also use our systems. So, you know
They're not in the way.
If if if if you meet Mitch by the Starbucks next to your office and Mitch talks about the cool thing that he's running, you know, you'll know about it. And it it takes a couple Mitches until certain people in certain levels that the companies know about it.
Yeah. They're like, I'll I'll give it a whirl too. No. I... Listen.
I I don't wanna say differently. I know people who work in different companies that are not using their products. So I don't think that's... I don't think that makes the product bad, to be perfectly honest with you. I think it just Exactly.
Yeah. They they they did the thing we all say we want is to, you know, for people to have a choice and and do what they think is gonna work for them Exactly. In their life. Yep. Yeah.
And I... Also, we have no idea. Like, I... Listen. I...
What I know about, you know, how how much I understand, you know, the world today and technology probably fits in a thimble. But I also would not be... I wouldn't be shocked if we woke five years from now and it was easy as, like, picking up your phone and going, make me a better insulin pump. And it... Like, maybe...
Like, I don't know what's gonna happen when recursive comes and when it starts teaching itself, and I have no idea. You don't either. Like, none of us know. So Nope. Yeah.
But, I mean, the... Just the idea that you could build a sandbox somewhere and just tell it, like, look. Here's my code. Just work on it. I'll I'll come back in in three months.
Like, tell me tell me what you come up with. Like, and maybe it'll... Maybe it would say, I don't know. It's pretty good. I don't see anything here.
And maybe it wouldn't. Like, maybe it'd be like, have you ever thought of this? You'd go, oh, holy hell. I didn't. But I don't think that's how it's gonna happen.
I think it's gonna be... If it doesn't have data, to your point, if it doesn't have people's graphs and how it intersects with how much insulin they got and where they were in their period when it happened, if it doesn't have that stuff, it's only so much it's gonna be able to do. Right? Like, it's it's it's math guessing up to a point.
We Are the Currency1:41:29
And That's actually that's actually really, really your point that the data that you were touching on, and and may may... Maybe on a note here. That's that's... Because we were talking about the ceiling, right, and limitations earlier. One of the one of the big issues actually for for us in the diabetes space is that, essentially, we are the currency.
Right? So, like, we are wearing all these devices, and these devices are collecting essentially our data. And for the most part, we don't have access to our own data. And to be fair, because we... You were talking about, you know, making making it available to to implement new stuff and to do research on new stuff and to, you know, to to innovate, we need we need access to this data.
So, you know, the right to my own patient data and the right to my own diabetes data is actually a pretty high commodity, I would say. And that is that is a real... It's funny that that we've, like, gone into this for so long now and only now touch on this, but, actually, it's one of them. Like, it's a really big issue, actually. I'm getting access to my own data because...
Again, not to, like, single out any any specific company, but it's just a good example. To to be to be able to go on Dexcom and to get a clarity report of, like, my last ninety days of, like, glucose readings is not the same thing as getting access to my raw data. I I hope that makes sense.
No. No. Of course it does. Yeah.
Yeah. And and so... And and there are many companies, almost all companies are actually not giving you this kind of direct data access. And as... At the end of the day, again, like, this is your data.
Like, you should own it. It was collected off of your body, and and you should you should be the the the the the person that can, you know, access it and can can do with it whatever for for for better or worse. And and, you know, use it for for research, use it use it to do your own analysis, or or not use it at all. Right? But it's one of the main things where these big companies are kind of, you know, putting putting stones in in the path here, and then we have to try and jump over those hoops.
And it's also... I... To to a certain point, how the the open source AID community and what we are doing with making these devices accessible, we're also kind of, like, freeing up access to people's own data by doing that.
Yeah. I mean, I... I'm... I find myself back in the same conversation from before because the the data is, I think, the most valuable thing they have. You you know?
Like, that's how they're going to try to make make money or value for their company moving forward. I mean, my transcripts are my data. Do you know what I mean? Like, everybody has some everybody has something. And I I...
Listen. I'm not taking a side on this. What I'm telling you is I don't know how you're gonna get it from them. I don't see why they'd wanna give it to you from their perspective. Do you know what I mean?
And from your perspective, yeah. Right on. You know what I mean? Like, I I completely agree with you. So I I don't know if it's gonna be one of those impassable things where we have to say there's...
I don't... I can't see a way around this. I mean, I mean, if you tell them, look. You can't have my data. They're gonna say, well, then you can't use my device.
And Well, see, and there's a compromise to be made, to be honest.
What is it? Like...
Well, the compromise could... I mean, just just a really easy one would be, hey. I'm... A, I am paying you money to use your device, so, you know, you've got money. B, I'm...
Like, you you you can still have the data, but I will also just like to have access to the data. Right? I mean... Because because, like, one of the compromises, and this is, I mean, not... Maybe not the best one, but one of the compromises could could be, for example, I am happy for you to also have my data, but I want to have access to my data.
Mhmm.
Yeah.
Right?
If you put yourself in their position, why would they say no to that, do you think?
Well, because we are the commodity. We're the commodity, and they don't wanna they don't wanna give it out. And then, you know... Yeah. Yeah.
I I mean, what would what would make you say... Be be...
Right right now, what's right now, what's happening, I mean, it it... It's a it's a crude example. But, like, right now, what's happening is, essentially, you're you're using Google Mail, and you just can't access your emails. You can only access a generated PDF of an email you've gotten through Google Mail.
Mhmm.
And then you go to Google and you go like, hey. No. I I would really like to access my own email, please. And they go like, yeah. No.
You you you get to use you get to use Gmail. You get to send emails. You get to read the created email read report, but you're not going to get your email. And that just doesn't make sense. Yeah.
And we should we should really fight hard for that and advocate for diabetes data rights and access to this kind of thing.
There was a big push around this a couple of years ago, wasn't there? But I don't know. Like, people... Because everybody was trying to... Or was it Tidepool trying to get everybody to...
It was for... Maybe it was for... Maybe I'm thinking of when Tidepool was trying to go to the FDA, they needed everybody to give them their, their loop data, right, or something like that.
Well well... Yeah. The... That... That's the the the big data donation project, and it was...
Parts of it were used, together with the data from the JAIP Center to, you know, put the loop algorithm through FDA approval. But, and that is still ongoing, actually. And then there's also, like, a second really big dataset where users of the OpenAPS algorithm donated their data. It got used in a four year research project here in Europe called the Open Project, and that data collection still exists. It's called the Open Data Commons.
But, yes, there are research projects where you can donate your diabetes data to a nonprofit organization, they will be doing research with it, which is which is one good thing. But, again, like, you can only do that, and you can only donate your data for research, for example, if you have access to your data.
Yeah.
It's kinda like, you know, you you donating your own your own body after you pass to research. Well, you can only do that if you have if you have bodily autonomy in
giving We say so. Yeah. Yeah. Well Yeah. I mean, to your point, I mean, I...
If I'm understanding this right, and that is the thing that was hitting me in my head that I couldn't remember what you just mentioned. But the only reason those people were able to donate that data is because they were using Loop and they had access to it. Correct. Yeah. Right.
So they were... Yeah. Yeah. Had they been using a regular system and said, oh, I would like to give... You know, I would like to donate the information about how this is...
There's no way that was gonna happen. Yeah. Okay. Alright. I get it.
Okay. Well, how do people get Trio? I know I... Listen. I'm just gonna come out and tell you.
Someone else helps me with Trio. I I don't know how to do it. But I... Maybe... Why don't we let you hear?
Instead of saying how to... So other people to teach me how to do it. Like, if I was gonna set it up today, how would I do it?
How You Actually Install It1:48:12
Well, so there... There's basically two two two building ways, I would say. So the the code is published on GitHub. It's it's a it's a platform where we where we host our code. It's it's freely available.
You have to register account on there, but it's it's free. You you can just get on there, and then there's the code, basically. And then because you're on iOS, you need an Apple developer license. So, essentially, you you register with Apple. You say, hey.
I am a developer. Unfortunately, you have to pay $99 a year, and then you have an Apple developer license. And that's more or less all... Everything that you need, and then you download you download the code, and then you can do two things. You can either use what we call the browser build.
So you go on GitHub and you use something that we as a as a community build out on GitHub where you enter six little text snippets, and then it'll automatically build for you once a week. It'll check if there's updates, and then it'll set... It'll send, basically, the built app to your phone in a little app called TestFlight. It's kinda like an unofficial app store, basically, on your phone, and you can go into TestFlight, and then you install Trio. Or you go the the more hardcore way.
I don't know. That's not real. It's it's not a hardcore way. I'm I'm trying to be to be jerky here. You can go on your MacBook.
You can download Xcode, which is which is an integrated development environment for iOS apps. And you can go on Xcode, and you can download the code from there. And then you can compile it on your MacBook, and then you can load it onto your iPhone. Those are, I guess, the two the two options that you have there. There's a bunch of things that the community has built around this.
So for example, if you don't really know how to use Xcode very well, you can paste a little... Like, you can copy a little text snippet, and you can paste it in the terminal window, and you press enter once. And then the terminal window will essentially do, like, a little guided conversation with you. It goes like, hey. What do you wanna build today?
And then you go one one is trio, and then you press, you know, next step, next step, next step. It it asks you a couple of questions. And at the end of it, it'll it'll basically tell you, if you press enter now, Xcode will open, wait until it has fully loaded, and then press the play button in, like, the top center of the window. And that's what you have to do. And then, you know, it has helped you through the...
What did I call before? The more Harker way? Yeah. It has helped you through the more Harker way, and it's also made that pretty easy for you. Most people that build nowadays are using what we call the browser build, so the one that is using these GitHub actions because of the big perk that I was kind of mentioning earlier is where it does an auto build every week, and it checks once a week, basically, whether there was an update.
And if there was an update, then it it skips that built. But once a month, it builds for you, so you... Your app never, like, expires, basically.
Yeah. Because I I can... I... Listen. I've lived through that a couple of times where my garden's like, my thing doesn't work anymore.
And I'm like, I I think that it it expired, I think. I'll text somebody. Hold on a sec. Happened once while I was traveling, by the way. And Oh, wow.
I had to text a person who had to contact... It was it was a
it was... It was a horror show. Actually, you set if you set up this... That browser built, like, you you could actually go on your iPhone and you can you can just press a couple buttons on your iPhone even and would be doing the build, and it takes maybe, like, fifteen, twenty minutes. And at that...
And then Arden has a new build on her phone.
I just texted the person who handles it for me. I said, we're using the browser build. Right? And they were like, yes. They also told me to tell you, oh, hold on a second.
Catherine said... She said, oh, here's a funny story that maybe Dan will laugh at. I once introduced him to somebody at Yale, and I called him Dan Moody thinking a 100% that was actually his name. And so I don't know if you if you clocked that,
but she apparently remembers it. No. No. No. No.
It it it was... So so, like, my my Facebook name is an alter ego. It's not my actual last name. And and she only had interacted with me on Discord where it's like... My my GitHub name is my disc...
My my Discord name. It's... Right? And then... And she interacted with me on Facebook Messenger.
And we had met in real life, and and everyone calls me Dan because I call myself Dan. It's my it's my... It's been my nickname for twenty five years, basically, since kindergarten. People called me Dan. And she knew me as Dan, and so she...
Like, she she introduced me to to this researcher out of Yale, and she went, so, yeah, this is this is Dan so and so. And then we were, like, texting back and forth, and at at some point, I I messaged her privately. I was like, you do realize that I am not I am not mister Moody. Right? That's not my name.
And and and she was like she was like, shut off. You you gotta be kidding. Yeah. She mortified in this text. Yeah.
So
Yeah. That's funny. No. I, listen. I...
So many great p... Again, so many great people. I even... It's not lost on me that I think there are there are people that are involved in this that we'll never know. I I...
I'll never know. Like, you know, I shared with you earlier, I think it would be awesome to just gather as many of them as you could together and let them tell their story about the part. Even if it's ten minute conversation or five minute conversation, put together a Yeah. You know, a... You know, an absolute, like, kind of...
I... I'd be so afraid to miss somebody. That really would be my concern doing that. Like, I would... I'd be so afraid to miss somebody.
Oh, I mean, we... So I mean, I... We we can, you know... Once we get off the air, we can probably set something up and and get you connected to a couple people. And I can I can only say the people that contribute to these to these projects, they live all over the world?
The developers live all over the world, and it's it's awesome that we do this, and it's awesome that we get together, you know, remotely and asynchronously. But in a in a cool way, there's nothing like getting together at, like, conferences and also at a thing like the hackathon that we'll be doing in Vancouver in November to, you know, get this really fucking super special energy in the room for, like, two, three people... Two, three people. Sorry. Two, three days.
You know, work on stuff together and just, you know, be in a room together with the people that you spend, like, dozens and hundreds of hours online with working together with. And Yeah. Yeah. And we can we can probably get you get you hooked up with a couple people if you wanna do this series.
Yeah. What would be interesting then is that, you know, as you talk to people, you would probably get the names of more people. It's a thing you could Oh, yeah. You could do ongoing. Like, I'm not...
Listen. I don't have the infrastructure to turn it into, like, you know, a documentary, but, like, it would be interesting to, like, let people share their remembrances of of how they were involved in it, you know, just to show the path. Where would you start? Do you start with do you start with lane, or is that not going back far enough?
You would you would probably start with, yeah, you would probably start with with with with Nightscout, I guess. Yeah. We will discuss... We will start with the Nightscout project, and, well, Lane was involved in there and a bunch of other people. I don't I don't wanna I don't wanna
alter the
names because I will forget names, so I'll I'll not name names. But Lane was one of them. And, actually, if you talk to Layne and if you ask him about, like, the origin story of, like, what he contributed to Nightscout and how that came to be is one of the most badass OG fun stories.
Yeah. Maybe that's maybe that's what I should do is start with him and then let him... Like, let the, like, the flower grow from there, and then I'll just keep working off.
And and then, yeah, there there are so many... There there are so many people in in a way, but but then again, at the same time, you would actually be amazed how Few people. How small the group of people is that builds this for tens of thousands of people and runs it and maintains it and keeps it alive. Yeah.
Yeah. I've I've I've... I'm sure I've shared this with you in the past. Like, I feel... Like, I've...
I have felt badly in the past because, first of all, it's not a world I know. Right? Like, it's, like, coding and, like, it's just not it's just not something I'm involved in. And, and then somebody comes along one day and is like, you should try Loop. And I try Loop, and I start talking about it from, like, my entry point into it.
And I I only know what I know when I start. And I've heard stories of, like, people have been insulted. You didn't say their name when this happened, or there's people that are involved. I'm like, I'm not trying to hurt anybody's feelings. Like, I think I try really hard to say there's a lot of people, to your point, maybe not as many as I'm imagining.
But, like, a lot of people put a ton of, like like, you know, sweat equity into this. They're never gonna make a dollar off of it. They're... They never... It wasn't even their intention.
Right? And they're... And they... A lot of their effort has gone into, like, a lot of the success that my daughter's had and a lot of other people have had. And I...
It's just hard that when you're talking, you can't stop to do your land acknowledgment every five minutes while you're trying to talk about something. You know what I mean? Like, at some point, you just kinda gotta let the conversation roll a little bit. But, you know, I've heard from people that they've been pissed in the past, and I'm I'm genuinely sorry about that. Like, I wish I knew who more of you are.
And this is gonna be... This is my attempt to, like, to do my best to put it on the record somewhere because maybe I have more of an ability to do that than other people do. So I... I'm hoping that'll... You know?
I don't know. I just... I hope people understand how all this is happening. It's... I think it's fantastic.
I've... I had the guy on one time who invented the Omnipod. I mean, it's just... That that thing is, you ever crack that thing open? Of course, have.
It's an insane machine. You you know what I mean? Like, that somebody, like, designed that one time is crazy. And then there it is. You know?
Like, and you talk to that guy and you realize, like, he doesn't work there anymore. He hasn't worked there in forever. It's just an interesting... Or I talked to the guy that put the assembly line together for the Omnipod five. Like, that's...
I think those are interesting conversations. You know? So I'm always trying to dig those people up wherever I can. Anyway, dude, we've been talking for two hours.
Yeah. We've been we've been on a long time.
You are?
Yeah. You were rushing home trying to get through traffic to do this
and now, like, you gotta be hungry by now.
I was I was stuck in traffic. Yeah.
Yeah. Well, just make sure make sure... I wanna make sure you feel like we talked about everything that you were hoping to talk about today.
I I think so. Yeah. Yeah. I think so. So, like, while we were on, I was actually messaging the the Knight Scott Foundation folks, and I was asking if there's if there's any...
Anything I should also mention. And they were they were kind enough to to to remind me that there there is, like, the recurring donation link. We were talking about the subscription thing earlier. Mhmm. And and and it would be ninescoutfoundation.org/donate.
Okay. And so if people wanna wanna give that, it's it's a really cool thing. You know? We we kinda need to support year round. The Ninescout Foundation funds, you know...
If if developers need devices to to work on these systems, they help fund that. And they help fund travel. They help fund these very cool events. And it's it's a nonprofit, and so we kinda we kinda need support year round or, like, they need support year round. I shouldn't say we because I'm I'm not involved in the foundation.
I I luckily benefit of the great work that they do. And, yeah, I think we've we've talked about the 03/2001 o release. It was... Actually, you know what? What I didn't mention back back when we were talking about this, and I kinda I kinda should probably...
Let me quickly, like, look up my notes. Go ahead.
While you're doing that, I'll just say to the Knight Scout Foundation, I hope you enjoy the either 500 or or $9,000 that I sent you this afternoon. Was If that much, could you send some of it back, please? Yeah. I
I I really I really hope we... Like, you didn't send your whole retirement. I know. I I I was joking. But we want you to we want you to have a good retirement here.
No. So the the 1 dot o release was... Represents more than a year of... Like, more than two years actually, but, like, the the core of the release represents more than a year of work by 20 plus people, and it ships a whopping 112 new features and fixes over 1,500 commits. So a commit is essentially like code.
Yeah. Like a a... How do you... How do I explain the commit? Jesus.
Okay. Pep... People who know what a commit is will know what a commit is. But it it... Ton of code.
And we added, yeah, four new sensors and a new pump to this. We we modernized Trio. It... It's got liquid glass added into it. We added a new alerting engine.
We added the cool quick pick quick pick treatment feature I was talking about that kind of learns based on your data, what you're doing, and gives you... You know, suggests treatments for you. And essentially, oh, I should have also mentioned that we invested a ton of time into making the app more accessible. So if you happen to be visually impaired and you have to use, you know, accessibility tools on your phone, the app should now also work better for you there. And I guess, all in all, I can only say I'm I'm...
I've been talking to you for two hours now, and I am a part of the Trio team. And I I invest a lot of my time there, but I'm not doing this alone. This is a a very cool team, like, you know, splattered across the world. And we're all doing this because we wanna help people. We wanna we wanna make diabetes suck a little less.
Two Grown Men in a Jazz Bar2:01:20
And if if if I can tell, like, a little small anecdote here and then maybe we can kind of end on that
Please.
A really, really good friend of mine... I I won't name names, but a really good friend of mine, we met at ADA. It just happened to be that he was also in New Orleans. And he has a kid, and his kid has diabetes, and his kid is using Trio. And we were sitting in a jazz bar, and he was asking me the same question that you were asking me earlier, like, why do you do this actually, dude?
And I was I was just saying, you know what? Between all the, you know, long nights and and sometimes also some of the drama that goes on the Facebook groups and, you know, and, you know, sorting through bug reports and and sitting sitting there being sleepless because you spent the last three nights on some kind of feature and something didn't work out or whatever. You know why why I do this at the end of the day? It's it's for for people like like your kids, right, who is is empowered enough to live a confident life and kind of can just go about about their day using Trio and doesn't really have to worry about diabetes because, you know, his his kid is so confident in in in their diabetes because of Trio that they just go about their day and don't really, you know, are worried that they're diabetic. And I said that to him, and then we kind of looked at each other, and then it was like two grown men sitting in a jazz bar in New Orleans kind of crying.
And, you know, ending on this note, like, this is this is why we do this. It's it's to make the lives of people better, because we live with it. We we are in their shoes. We know what it is like, and we just, you know, wanna wanna make diabetes suck a little less.
I I mean, I can't I can't agree with you anymore. Anytime I hear from somebody and they're like, it doesn't matter what they say. It it it really is. It's like something got better for me, and it it had something to do with something you touched, and I think, well, that's that's really fantastic. I wish I wish more people could put their their energy into things like that.
I feel lucky to be able to, to be involved in something that somebody, you know, says is helpful to them. And you guys are doing it in spades, but it's also very behind the scenes. I mean, it just it just really is. Like, you know what I mean? Like, you don't...
If you made a podcast about this, like, six people who understand coding would listen to it, and it'd be really interesting to those people. And and you you... You know, how do you get out in front of it? Or, you said, you try to, like... I don't know.
It's just... It's a hard thing to put a face to, and I'm... I hope I'm doing that to some degree. So Yep. Yeah.
I would like I would like for you guys to get your flowers and, and to feel as... To feel as the gratitude that I know everyone feels. They just don't know where to send it to. You you know? Like like, if I'm just a regular person who downloaded Trio, I I mean, I imagine there's people who did it, but I don't know the rest of that story.
Oh, actually actually, there there there there are there are more and more people doing that. We've we've we've the only for five release and even before, we we we are seeing such a massive influx in in, like, new users. And the great thing about Trio is that it has, like, this very detailed onboarding guys. So usually, they they come somewhat equipped and then, you know, and then they come across the the open source community and they come across the the various Facebook groups. Because, you know, the the Trio the Trio community has a similar group to to the the Juicebox podcast group on Facebook, for example.
We're not as big because, dude, your group is fucking huge, but there's there's still a couple of thousands of people in there. And they can ask questions, they're helped. And and then they they learn the the system, and then they use it, and they love it.
It's amazing. Yeah. And I'm glad I'm glad people are reaching out and saying thank you because, I mean Yep. Obviously, you're helping them in ways that, mean, it's it's just hard to put into context how how exactly valuable it is. It's it's...
You know, there's nothing but... There's nothing more important, honestly. So... Alright. Jens, thank you so much for doing this.
I really do appreciate it. I... I'm... We're really gonna be testing our new system about how we track curse words and remove them from... Because you're a...
You you used a few today. So this
I'm very sorry.
No. Don't be sorry to me. It's a a lovely girl named Catherine who's gonna be handling it. So... And different Catherine that I spoke about earlier.
So good luck, Catherine. Go go get them. Yeah.
Sorry, Kenny. Oh, I'm very sorry.
Alright. Hold on one sec. Something new just arrived from MiniMed. It's the MiniMed Flex system. Think about everything people love about MiniMed technology, designed for great outcomes, helping keep you in range, and made to be easy to use.
Now in a pump, it's the size of two stacked vials of insulin. MiniMed Flex is fully controlled from an app, designed to stay tucked away, empowered by the proven automation that helps over 600,000 people manage diabetes day and night. Order today, minimed.com/juicebox. Having an easy to use and accurate blood glucose meter is just one click away. Contournext.com/juicebox.
That's right. Today's episode was sponsored by the Contour Next Gen blood glucose meter. It's the meter my daughter uses. She has it on her right now, and you're one click away from having it too. And don't forget using juice box podcast links helps support the show.
And don't forget when you're at cozyearth.com and you put all that great stuff in your cart, you need to use the offer code juice box at checkout so that you can save even more money than you were going to. Cozyearth.com. Use the offer code juice box at checkout. Get yourself a towel. Listen.
I'll tell you right now. I'd get the towels. I'd get the sheets. I'd get the joggers. Probably a t shirt, definitely the blanket for the sofa.
These are things I would do. You do what you want. And while I have your attention, nobody's asking me to say this, but I just wanna add it here. The Night Scout Foundation could use your help. Nightscoutfoundation.org.
Donate to the annual hackathon. They're up to 59,000. As of this recording, they're trying to get the $75,000 to get all of the great people to the event so they can work on the next thing the DIY community will be bringing to you for absolutely free. It's not bad. If loop or trio or something like that has benefited you in the past, throw them a couple bucks.
Friends, my diabetes pro tip series is about cutting through the clutter of diabetes management to give you the straightforward practical insights that truly make a difference. This series is all about mastering the fundamentals, whether it's the basics of insulin, dosing adjustments, or everyday management strategies that will empower you to take control. I am joined by Jenny Smith, who is a diabetes educator with over thirty five years of personal experience, and we break down the complex concepts into simple actionable tips. The diabetes pro tip series runs between episode one thousand and one thousand twenty five in your podcast player, or you can listen to it at juiceboxpodcast.com by going up in the menu. If this is your first time listening to the Juice Box podcast and you'd like to hear more, download Apple Podcasts or Spotify, really any audio app at all.
Look for the Juicebox Podcast and follow or subscribe. We put out new content every day that you'll enjoy. Are you tired of scrolling past clips that last five seconds? Are you an adult? Would you like to have a long conversation?
Juicebox podcast is the place to be. Follow and subscribe. I'll be here tomorrow with another episode. I hope to see you again.
Read the full disclaimer
© 2007–2026 Juicebox Podcast. All rights reserved.
