In part two of our series with Umang Patel, CCIO at Microsoft, Tom and Umang explore what the NHS's digital future really looks like once you get past the infrastructure excuses - and who's actually driving change.
Drawing on real clinical encounters, startup lessons, and a candid conversation about purpose vs profit, this episode argues that the bottleneck was never the technology. It was always the context locked inside people's heads - and whether we're brave enough to let go of it.
Tom: One of the things I'm really interested in in this new AI enabled world is the value of context. There was always a view that your data was your moat as a business, and to an extent it still is. But I think, you know, if you listen to the latest podcasts - of course, this is one - people are saying that data is less of a moat than it used to be and certainly features on the platform aren't. But the bit that is really important is context. So the micro decisions that are made.
So I've got an example from a client who I won't name where we were looking at booking in cardiology. And the general manager said to me, the problem I've got is I've got three or four people that book my cardiology appointments and only one of them books Tavis. And if she's not here, no one else will do that job because so much context is in her head. So she's on holiday. We don't book any Tavis for a week or 10 days. And it puts my whole operational thing back. And being able to capture that context in an agent. And that's obviously a relatively simple administrative example, but taking that into the clinical space and then, you know, cross-functional teams. That feels like the opportunity the NHS has to build something which is really amazing and special.
Umang: Yeah. Actually, you say that because the other thing, I could build as many agents as you like, but you still need somebody to deploy it and go off. And I think we have that like nowhere else.
Tom: But I love these little examples because - and I think maybe that's a difference. So if we sort of wind back just after COVID - I think just before COVID when we thought we were winning - it felt like we were going to be able to do anything and we're going to by hook or by crook find the solution to some of these things, except for maybe some of the bigger infrastructure things that we just hoped would come in time. And we, you know, that was a piece of string. So like, how long until we prove that enough people have mobile phones or data or whatever those bits were, or security was there and we weren't quite sure about - but the rest of it was easy. That's sort of gone like 180, hasn't it, where now it's like all of the infrastructure stuff seems to be improving.
Umang: Yes.
Tom: But the little ideas that you can build on - but in that example, in fact, today we're talking about MDT coordinator and being like, I've got the most amazing one. But truthfully, when they retire - which is coming up, right - like no one wants that job because it's not a job. And I was like, well, it isn't even funded.
Umang: Well, no, actually, MDT coordinators aren't funded - yeah, that's interesting. And the first thing you do, so you end up getting - well, then the only person that becomes booking it is your clinician.
Tom: Yeah.
Umang: And you're like, well, you don't want - you want them doing their job, but that's the way it goes. So I think what people are saying is like, okay, well, I can want to scale that. I love the idea of people being able to like, monitor their expertise. So I'd love to wait for people like that person to be like, okay, I'm going to write it all down. And every time someone uses it, you know, 20p is going to go to the hospital charity, right? Or whatever it is. Like that sort of thing - and then you get into this whole exciting space of abundance where you go, well, now actually, I can book Tavis all day long. That's not my rate limiting step. What's the next thing that I need to do that's going to improve the population of the group? Is it I need to do this at a more regional level? Is it that I need to link that up to the next part of the service? And I think, you know, it's those building blocks that we're going to start to see more and more.
Tom: A hundred percent. And it's, again, the management comes out - it's looking for the bottleneck in the system. So where's the bottleneck? And when we started Dr. Doctor, a lot of what we were trying to do was - yeah, the end goal was to solve supply and demand. That's what we wanted to do. But it said that on some of our pitch decks, and actually what we realized was in 2012, the problem was getting just basic information to the hands of the patient - like just simple stuff. And the businesses that forgot about that step and went straight to trying to do something higher up the value chain really struggled. So we intentionally started with the basics that gave us a base of patients, which we could then build on over time. And so I've always believed in that kind of stepwise approach - which is why now is so exciting because as you say, the ability to take in our case, billions of rows of data and hundreds of millions of patients, and then suddenly turn all of those into a whole new set of value is massive. And so maybe actually, when we talked about the journey, maybe it's actually about - we're close to the tipping point and it's going to take a few people to just push there. Um, what do you push down hills? I don't know. Snowballs. Elephants.
Umang: Yeah. Maybe like characterizing it - I think maybe when we first met, it felt like we were building up to a crescendo. Right. Like I was trying to get to an end point - like a story arc that was, you know, beginning, heartbeats, overcome the end point. And of course, I think what we've realized now, past COVID and sort of all of the other stuff that's going on - like, even if you think you are going to get close to an end point, that is just like a new beginning of another thing.
I had this the other day - so I'm just going to blatantly steal it - but like, I think it's more about life. But anyway, the whole point was like, rather than trying to think about it like a story where your final page is the most important thing, you have to sort of think about it like a song. Right. So you're not waiting for the final note, you're trying to see like what mood it puts you in. And what I would love us to be like - what chorus are we trying to write? There's something about us going, like, what chorus is it that we like? And that sets the culture, you know - what are we going to do that's going to become that repeatable bit that gets us from - yeah, well, that would be it.
Tom: I mean, we should definitely write the chorus together. I love that. I'm sure there'll be lots of people listening in. But do you know what's interesting, actually, because the startup world is so obsessed by next funding round, next funding round, the exit. When's the exit going to happen? And it's contrary to what you just said, but actually the businesses and people that have managed to create some success in health tech at least are the ones who I think have had either a longer term goal or have done it because they're enjoying singing the song. You know, and they're enjoying the journey and they're enjoying the purpose. And maybe there really is something in that because none of us are going to fix medicine in our lifetimes. Right. It ain't going to happen. But I think all of us can contribute to an orchestra of change.
Umang: Yeah. Oh, nice. OK, good. We're going to put the band together. And I think that is true. I think old VC funding models was like, no, no - you're a lead singer and that's it. Right. Like you're Lenny and everyone plays to your beat. You can't. Whereas now I think it is much more - and hang on a minute - we're going back to people being driven by purpose again. I remember, just by way of story - Dave Lawton, the chief exec at Wolverhampton. I remember like during and after Covid, we're doing lots of stuff there with them. There's Babylon and he's phenomenal. Actually, you should get him on. He'd be amazing. But anyway, I remember him being like, the problem I've now got is if I wanted to tackle a backlog or do something, I knew exactly what everyone was doing. The reason that I was around was - he's like, well, I knew who just, you know, got divorced and had a second life and needed more money. He had like, he just knew all of these things - which first is just part of knowing people, right.
He's like, all of that - like the soft bits. So the problem is when Covid happened and we saw getting through it, I'd be saying to people like, work harder, do more. And they'd be like, but why? I'll pay you more. But I don't - I want more time with my kids. I want time away - I don't want to be tied to coming into hospital. I want to be able to drop off pickups or whatever else that might be. And I think the unifying thing, which we're now learning from it, is - less people I would hope want to become billionaires, but more and more people want to be part of that community of purpose and change. And I think that's what we're seeing in hospital. So, you know, if you try and go to people like, well, I'll just pay you more to fix a backlog - I'm sure you'll get some hands up, but nowhere near as many as you would have done before. Whereas now - do you want to be part of building this new thing that's going to make healthcare better for the next generation? Do you want to build an ABT solution or think about population health in a different way? Like a lot more hands have gone up.
Tom: Yeah. That's really cool, isn't it? It's certainly true that the social contract between doctors and the healthcare system was beginning to get broken anyway. And then COVID did just completely shift it. Not just for doctors actually, for everyone. And I also think - I have this slight obsession with the fact that remote working - Dr. Doctor is, we mostly work remotely, but we intentionally get together occasionally - creates a more mercenary set up as well, because you don't know the people in the same way and you need other things to bring them together. I suspect the other thing is that, you know, you and I - let's be honest, we're kind of nerds, right? Deep down. Or quite obviously so. And we had those early colour phone, mobile phones - or, you know, we saw that these technologies were going to change the world. And we were maybe early, maybe we were a little bit ahead of ourselves. Whereas if you're a doctor who's just studied medicine in 2026, you're about to become an F1 - you've lived your life through technology. And it must be bizarre, I think, looking in and thinking none of this stuff's being used yet. So the intrinsic motivation driven by purpose and the extrinsic motivation driven by - well, the rest of the world's doing this, why aren't we - I think they're a powerful and potent combo.
Umang: Yeah. Well, I mean, like - you know, all these pictures - I remember having a Psion followed by a Palm Pilot, followed by - I've got all the books we used to have to carry around on year one, I'm completely aging myself. But like - and then the idea that they were going to - even the BNF, like, okay, great. Now, like, it took seven years between us saying there's a BNF available online to it becoming anywhere near useful on some form of mobile device, which wasn't your phone. It was like a Palm Pilot that had a separate thing. And say, yeah, I think maybe back then I should have thought about it. It was hardware, right? It was catching up. And then we suddenly got like faster processing, iPhones, etc, etc, that sort of became ubiquitous from the hardware side. So yeah, maybe this whole new generational thing going through is going, well - the hardware's there now, like, why hasn't the software caught up? And I think that's a problem.
Tom: That is a problem. Yeah. I was chairing a roundtable for people with or who had survived cancer a couple of months ago. And a lot of people were showing their things. And one of them was like, why can't you just tell me that I've been spoken about on the MDT? Like, I have to chase my GP. Like, why can't I get the blue ticks that WhatsApp has? Like, I've been discussed. And therefore the next step is happening. And I think that's a software problem, not a hardware problem. They weren't sat there thinking - I have to stay near my phone, or are you going to send me a letter so I can get it - or whatever. They were like, but I've got the hardware and you've got the hardware - like, fix the software issue.
Umang: I mean, yeah, again, I think that's a real opportunity. I mean, there's about 10 things in there that I want to pick apart. I think you're absolutely dead right about the hardware-software thing flipping. I love the idea of the blue ticks. I mean, hopefully the product team are listening - let's go code just to get that done. That would be amazing. There was an amazing guy who sadly has passed away now called Michael Serres, who had Crohn's. And I was early on in my journey. He was innovating around wearable devices. He had a device to help sort of dictate when his bag had been emptied and stuff. And I remember him standing up at conferences, and he would always say, I have asked my consultant to tweet me my blood test results. Because I don't care about the idea around this. I just want to know what they are. And I don't care who else knows, just let me know them as soon as possible. That sort of championing was needed in those days. But the idea is the event happens, and you get instantaneous feedback. Why shouldn't we live in that world? I think that's a really exciting future to try and prescribe.
Tom: And so let's dig into this software point. So the software is lagging - 2025 December, we get Codex 5.2, we get Opus 4.2, if I got those right - I think I've got those right. We've got these two new models that suddenly shift the balance from it's interesting using AI to help me code to actually we can do full agentic coding and the quality is really high. And it felt like - for me, it was over that Christmas period - I went from, I always do a little tech project over Christmas, and I downloaded Antigravity, which is Google's fork of VS Code - allows you to use Gemini sort of as an agentic coding harness. And I was amazed by how quickly I could go. I built a bunch of stuff. Luckily, none of it made it to production, don't worry - but at a rate which played in my mind. So that was like, for me, a real moment. And I think while I was doing that, the rest of the world had that same moment, right? We went from the start of December 2025, to where we are today. And all anyone's talking about is agentic coding. So suddenly, as a junior doctor who's got the hardware and it feels like the software isn't keeping up, you can make your own app. And like, what does that mean for all of us who've been in health tech for all this time? What's it mean for the NHS? And, although I don't love to talk about regulation, what does it mean from a regulation point of view? And how do we find the right balance between pace of innovation and making sure it's safe?
Umang: Yeah, yeah. A lot. But we do think about it all the time. I think that is - I like how, again, thinking of an analogy on the fly - it's DIY. In the truest sense of that word, isn't it? Like, all of a sudden, you can now mix any colour paint you want, or whatever it is - or, you know, I can't think of, I have to ask a builder - what was the thing that was really hard to do? And then all of a sudden, a drill came along, and you get that bit or whatever else it was. And then I definitely think there's - it's probably something like the wall plug, isn't it? It's something along those lines of being like, actually, it was really hard before, and then somebody just did something that made it pretty simple. I think some things will go wrong. Of course, like, hanging a picture in the wrong place is not as dangerous as what might go wrong in healthcare. So I think we have to set some guardrails. Again, I think that needs real clinical leadership - like, what are we willing to accept or not? And I was sort of joking about this earlier - I was like, look, it's a bit like saying your sat nav - if every time it says turn left, it doesn't say, like, oh, by the way, remember that you're driving. Like, it assumes you're a responsible driver enough to at least be reminded at the beginning. And then you get on and do the job together.
Tom: And I think that's a lovely point to be able to do that. I think the same applies - I'm going to trust you as somebody that provides care. So I'm therefore going to let you do it and not remind you everywhere. I might remind you upfront, because I should, because this stuff is net new. And I will audit you off the back of it. And like, we should try and make it as safe as possible. But this continual reminder seems like it's stifling any of the UX or usability. It is interesting. The sat nav example, because you do still hear stories about delivery drivers that have driven into the sea, because they've just followed the sat nav. So it does happen. Right. Yeah. But I guess, every single day, 99.999% of people managed to get to their destination safely. So we are going to have moments where things happen, and they'll probably make the news, just like it does when a DPD driver drives into the sea.
Umang: Well, I think the difference for us - and not that on the sat nav point - is say that patients are like, well, hang on, I'll just drive. Right. Like, all of a sudden, you're like, well, hang on a minute - I haven't. Like, what if I - someone else will - if I'm not going to get you to where you want to be, then, you know, I'll have to find a way by myself. And we're increasingly seeing that. It is - it's happened now more than once where somebody's come in and very politely said, look, I'm not really here for your opinion, doctor. Like, I'm here for - when can you book me for the MRI scan? I've done all the research I need to myself, and that's the next bit and you're the conduit to the next bit. And so I think there's something about that's hard for people - for some people - as a clinician to change the way they sit in their seat, having someone possibly knowing more about their rare disease than they do. And like, how do you think that sits with some people?
Tom: I don't know. Well, it's happened to me, so I don't give anything away, but there was a child that came in and the mum of the child Google searched the treatment plan. It was a relatively rare condition - an incredibly rare condition - and like, to a point you had to phone Great Ormond Street up to get the specialist advice and all of the care was safe. But anyway, the whole point was - the week before the baby had a scan and a load of tests done - come in with something slightly different and they were like, well, what about a scan and all the tests? And we're like, well, we can't do them every time because we've got to see how this evolves. And like, we're not against doing them. By the way, we're not against it - it's not like I'm trying to protect resources here. It's rare enough that we'll just - but you know, we get to. And I said, whilst you're waiting to see how this next symptom evolves, like you're free to go home and we've got open access, which means you can come back anytime from the hospital. Because we're just watching and waiting. And I sort of went back in the room about half an hour later and they hadn't gone and they were really a little miffed, right? And the parents were like, look, we don't want to go home. Like, why was it sending you home? It was saying, I just can't. And then the mum turned her phone around and showed me that she'd been Google searching it. And it had gone on to an AI assistant - I don't think she'd done it intentionally. And the AI assistant - Gemma, I think - had said - in the thing you've described, and she'd actually read it all, like she prompted it effectively going, well, I've got this. And it said, well, the next thing to do is get more information from blood tests and a scan. And so I was sort of saying - well, I don't know how to - I was like, look, that's not - you know, when somebody's got that as an anchor point, it's like, I don't quite know. And then she said to me, so brilliant - she's like, well, would it hurt? And I was like, well, no, I suppose it wouldn't hurt. And then I was like, what do I do? Do I get my AI chatbot out? Like, do I get Open Evidence or something? Like, I hadn't really thought about it anyway. Just because at the end of the story - things evolved a little bit during the day. We did end up doing the scan and that family looked at me as being the block to have it happening earlier.
Umang: How interesting.
Tom: And so the dad said to me, as I walked out, he was like, see doctor - every day is a school day. And I remember sort of going out thinking like, oh, I haven't quite processed that. But there was something in that. I think it was really different. And I was like, actually, they did absolutely what - like, if I would have supported advocating for a child, absolutely using all of the tooling available. We were ill-equipped to be able to come back with it. I think we do have to change our priorities. I was like, how do I train myself on that? Or how do I train anyone else on that? Like, I'm pretty geeky, pretty articulate and pretty senior at this point.
Umang: So do you know what - it completely freed me, of course. I mean, they were lucky to get you to be honest, because not only have you taken that and obviously reflected on it - it might actually systemically change the system. The answer is not to be like, don't use Google, or don't use a chatbot.
Tom: Nor was it that I think the LLM had done anything wrong, right? So this wasn't the LLM saying, oh, by the way, you need to drink pickle juice or something - it was quite sensibly trying to put things through. And I think we therefore need to have better ways of communicating. I think we talked about this - going, well, I wish I had some ability for them to - if I had sent them home - maybe going, well, actually, this is going to happen a lot. Here's a way of you recording all of the symptoms, and pressing a button that gets an instant phone call back or whatever - we haven't worked it out, but yeah, I think we're going to have to change.
Umang: I think that is such an important point, and these things being viewed as a genuine change to clinical practice - and not an IT project - is going to be critical to their success, because they're not IT projects. I mean, they are partially because you've got to get the tech working, and you've got to have a microphone that works, and all these sorts of basics, but assuming we can solve that - and we can, it's 2026 - it's a clinical change project. And that's why it needs the clinical leadership that we discussed earlier.
We are on the lookout for industry leaders to join our podcast, so if you want to be part of the incredible conversations we are having around digital transformation within the NHS, we'd love to host you on DrDoctor Will See You Now!
