E641 | This Week in European Tech with Dan, Mads, Lomax & Andrew – AI, Robots & Regulation

28:40And bad stuff can happen from that. You know, like if you go back to history, you can be like we had revolutions off the back of the Industrial Revolution. The welfare state didn't kick in to sort of modify the effects of this until, you know, after the Second World War. So all of these things that ended up being sort of necessary to, I would say, rebalance society after what happened with the Industrial Revolution, it kind of needs to happen again. We talk about universal income, these kind of things. So I'm kind of glad actually. So going back to John Thornhill's article, the real industrial revolution is happening in China at the moment.

29:17The US, Europe is third, definitely in that tier. It's a very slow burn if you're looking at it as an industrial revolution, but sure. John Thornhill Revolution is a romantic word. It sort of conjures up fast, speedy, impactful change. Yeah, you're right. One of the problems is that we started with a blank sheet to some degree. If you think about the key pillars of the original industrial revolution, it was energy, basically transition. So from animal and water power to coal and steam, it was transportation from horseback to canals, railways, steamships. And it was, I suppose, sort of institutional.

29:59So think about things that were put in place that allowed the whole shebang to accelerate. So patent laws, banking system, trade associations, organized capital. It's a bit of a tired trope to say industrial revolution now because first of all, is it a sort of 10x, 100x better solution? and AI in niche sort of applications arguably is, but it's not that foundational yet. It's not that foundational sort of sweeping change where you go from a horse to a train until it's deployed in the organization. Yeah, overnight, exactly. And that's the key because you're not working with a blank sheet. You've got a huge amount of friction to deploying this stuff and getting it working.

30:46And you've got a huge amount of friction to making the changes and unwinding a lot of the institutional, regulatory, cultural issues to change from how we do things today, which is already complex, to doing things far better, but in a completely different way. And I think that's where the analogy falls down a little bit. So ultimately looking back, I think we'll look back and go, yes, AI was a revolution. What the inflection point will be, is it now the start? Or is this now the efficient? I can't remember the steam engines. There was a steam engine, but it was highly inefficient. and then someone fixed it.

31:19I can't, my history doesn't serve me well. But it took a while before it actually really ran away. It was the rocket, wasn't it? Wasn't it the first one? Well, that was the train, yeah. But there was a problem with it. There was a problem with the steam engines. I think you're absolutely right. There's going to be lots of friction. There's going to be lots of issues. But the challenge is coming back to some of these luxury beliefs is that we have time to, you know, really give them, you know, decades to play out. I think if what we do is lean back and say, look, we're going to have a long time to figure this out.

31:51You know, steady as she goes, China is going to race so fast. We're going to be so far behind that we're going to have a completely different conversation. Imagine a world where Europe can no longer make any automotive that's competitive in any category. What will that do for societal upheaval when all the German automotive companies are bankrupt, when the whole automotive supply chain is gone? Right. And so I think to some extent we're sitting around. We're talking about how we need to do this gradually and gently so that we, you know, not people don't, you know, and all this stuff. Yes, that's true.

32:20The challenge I have with it is we're now living in a context where there is a real competitor. It's not like the UK where there was no real competition and the UK could take a little bit of time to figure this out. The competition is there and they're running fast. We just learned Mandarin. I don't think people don't feel the fire. um interestingly we're chatting to one of um this is slightly tangent but it's the same point biotech right if you look at um if you take biotech as an example now there's a general consensus now so u.s has always been a leader generally in biotech right is it a large consensus from chatting to a biotech founder in our portfolio last week spending quality time with him have he just spent time in the u.s is there's a consensus amongst the u.s and eric schmidt put out an article in Time about six months ago about this, that actually China is the leading, China is leading in biotech now.

33:14Having gone from nowhere in like seven years, is now the leading like in terms of new therapies, in terms of new tools. And so in terms of CROs, like all the big CROs now are generally there. So I think just that's a small example of like a really important foundational technology where China is just, we can navel gaze, but China is just cracking on ahead. i think you can probably you can argue that for many sectors like max well i mean we just talked about in the context of ai and robotics yes but i know about core like foundational sectors yeah yeah i think i think i think i think also we just don't see or the the western media doesn't show what's going on on the ground in china so i just i just don't think it's in our faces but my senses they are they've i think like mad is right like we don't people people in europe don't feel the fire underneath them at all.

34:06And so it's going to be a bit of a wake up call. I mean, to some extent, the age old discussion between employers and trade unions is fine, right? You're discussing how to share the cake, right? There's a nice big cake. How do we share it? And my worry is that as we sit around and pontificating on how to share the cake and how can we best look after everybody, the cake is going to be gone. Yeah, the Chinese will have eaten it. There won't be any cake. Have your cake and eat it. Because there is a way to share productivity gains with employees. It doesn't all need to accrue to the equity holders, you know, there isn't a way to make this work, you know.

34:39Matt is touching on the key point, which is the original industrial revolution was mechanizing muscle. And now we're really mechanizing minds and the minds are in the heads of people. And so one of the reasons China can do things so quickly is because they ride roughshod over the opinions of many, you know, if the great red state decides it wants to build an airport, it does it. And so when we're thinking about how quickly we can keep up, precisely to Matt's point, if we're still arguing over how the cake is shared and what we shouldn't do and who it's going to annoy and Lomax's favorite NIMBYs, then I mean, that's the friction here.

35:18It's not technological capability or capacity. It's the societal ability to absorb the change and for people to make unpopular decisions or unpopular with some people so that we can we do there is a care of duty as well we do we do have to i don't know how you do it but we do have to find a balance which is also looking after people i think it's important um so the longer longer term this but you can turn that around and say the longer term you're you're neglecting care by trying to pretend the change isn't happening like you're not stopping ai and and it'll be you know if you get You can either be in control of the pain and manage it proactively and then benefit from not being left behind.

36:06Or you can be reactive and be left behind and then the pain's worse and you've still got to deal with all the same pain anyway. You need technologically literate politicians because actually, they don't understand. No, not technocrats. We haven't even got business literate politicians. Let's get some commercially Lomax is right Actually go back to the industrial revolution You had the communist revolutions In the wake of that And actually you know what If we fuck this up If we fuck up AI and robotics You do end up with social upheaval You do This is why politicians Actually hate AI Some of them do But they're putting their head in the sand you need people who actually can see across the big picture and to navigate this because it's a big it's a big change there's a great example of this which is um estonia which is thomas hendrik i mean before he was president he was a software programmer and he is the reason why estonia has now got you know it's fully digitized it's got you know you can be a digital estonian citizen one of the fastest growing economies yeah exactly because he had he had a digital vision he had a Tiger League project sorry so when people understand you can't do this stuff and be visionary about it if you don't understand it to Lomax's point well let's lean into what the UK is doing so the UK this week has launched a couple of initiatives one which is this AI sandbox which is working out how to I don't know how we really unpack this but it's from Liz Kendall wasn't it she was at the the Times Tech Summit this week and she's the technology secretary I think also the NP for Leeds I think but anyway somebody can correct me on that so she's announced plans to look at how established companies and startups can test new AI products in the real world some rules and regulations will be temporarily relaxed and I got into a big fat debate with some investors and founders this week which was basically them against me.

38:19And one of their kind of framings was, well, this is just like the FCA sandbox, which actually led to a lot of good things where startups and businesses could play in a relatively free environment to get new financial services to market quicker. And I was saying, well, this isn't financial services. This isn't fintech. This is AI. And also the UK isn't the same place as it was 10 years ago. This is a very different realm, a very different financial and commercial situation we're in, I think. So I was saying pretty much along the lines of what we've been talking about, we need to approach this very differently.

38:56So am I right? Am I wrong? How could this work? What should the outcomes be? Where do we focus? Mads, kick this off with maybe looking at the FCA sandbox and what that was. Yeah, absolutely. So the FCA sandbox worked because when you launch a financial services company or a fintech, you need authorization to operate. So you can't launch a challenger bank or a payment service or any of these fintechs without regulatory approval. And then the sandbox provided a faster pathway to get the regulatory approval and sort of solve some of these gatekeeping issues to make it easier to enter into experiment, which was great because that's what you need in sort of a highly regulated industry.

39:37Now, AI as an industry isn't gatekept. so that, you know, meta, Google Anthropic, they're all deploying AI at massive scale with no special permissions. You don't need an authority to authorize you to deploy AI. So that begs the question, what barriers are actually being removed? And, you know, I think the key place you can look to is the NHS. You've got an absolutely massive diagnostic backlog there. You've got more than a million and a half people waiting for diagnostic tests. and you have almost 400 ,000 people that are waiting for more than six weeks. Now, the standard or the target is to get down to less than 1 % of people waiting for that long, but it's now almost a quarter of people that wait for more than a month and a half.

40:25And so one of the things we've talked about before here is, could we make a healthcare opt-in model work where you said to people, look, maybe this technology hasn't been tested for three years, But if you opt in, we're going to give you an experimental diagnostic. And if there is an issue, if something props up, we're going to have a kind of a radiologist check it and make sure that it's okay. And that could be a way to deploy AI more quickly to address a specific issue around wait lists and give people some flexibility and sort of not lowering the regulatory hurdle. You're still going to go through a process, but making sure we can deploy AI more quickly.

41:05So it would be a very, very different thing from what we saw at the FCA sandbox, but it could potentially be something that could work. That was something that I suggested in the debate that I was having is I think this is a great thing if it was ring fenced even further. And I use the NHS as an example, because I get that also government and officials need to see the real world operating so they know and have the insight into how that how that private market is going to behave in the real world. So I kind of get it from that perspective as well. but Lomax what would you what would you add or change or debate no I think it I think it's I think it's good as in it's a it's a it's not an it's not a straight analogy from the FCA sandbox but it is a sandbox of sorts of like if there is a streamlined way with a in a way it's a more of a streamlining of procurement processes which it was different for FCA sandbox was a sort of simplification of the regulatory rules.

42:06This is a procurement, a streamlining and simplification of the normally onerous government procurement. But that's only on one side of the equation, right? So this is only how government can buy and operate. So that's another way of ring-fencing. Anyway, I interrupted you, so you carry on. Just for example, if you look at the NHS, for example, which is basically 20 % of government spending and is a large proportion of GDP is how can we, if we want to maybe have two aims, let's say, the one aim is to create homegrown technology champions from the UK and at the same time is have the NHS benefit from this technology, then the sandbox in this case is, I think, lighter.

43:00you could have a lighter regulatory burden, but that can be quite tricky in healthcare. But you certainly can have a streamlined procurement process and be like, look, we're going to test this out in this subset of NHS trusts. And I actually, with Mads, is what Mads is saying is I personally think my instinct is the sort of low hanging fruit with LLMs in particular is around the administration of healthcare, not the actual diagnosing and treating of patients. Because you talk about AI radiologists, whatever, that's actually not LLMs, but it's just like, it's, it's, that's also hard because then you're getting into like giving clinical, you know, it's like, that actually for me needs to be a relatively high, I'm not going to, I wouldn't, I personally wouldn't be shortcutting too much the regulatory process for those products.

43:49Although in the US you have it, it's called the therapeutic use exemption so in the u.s where you have um an area you have an area of particular acute need with like lack of um current standard of care in the u.s you can get fast tracked so we could have something like that which we probably already have in the u.s you'll probably know you'll probably know who this company is i saw i'm sure it was this week a diagnostics startup had raised I think in the hundreds, maybe 200 million Series B. Good for them. To do, to effectively, what sounded like becoming, it had absorbed all of the doctors, journals, all of the information.

44:30And I cannot from the life of me remember what the name of the company was, but it felt like that was not just on the operational side, it was actually client facing diagnostics. Do you remember who that was? Did you see that? I don't recall. There are a few companies doing that, But yeah, it kind of consolidates your EHR, et cetera. But I personally think it's even just like cutting waiting list times through administration of healthcare. It's like this sort of you can predict if someone, for example, this is very simple stuff, but, you know, you can use AI to predict how likely someone is to miss their appointment, for example, depending on what time the appointment is, how far they live from the hospital.

45:04And like putting in place processes to make sure that, you know, they get a series of reminders, et cetera. Can I ask you a question, Amos? Tell me. Do you need an AI sandbox to implement that? So, okay, I would call it a sandbox because normally to sell that kind of product, which is probably a medical device, into the NHS, it takes years. So I would create a sandbox. What do you call it? An AI sandbox. I don't know. but it's a sandbox to effectively allow these products to have access and get real use case get the products into the hands of people as soon as possible for sure then so we've just come down from an ai sandbox kind of more general to an nhs one to a procurement one within the nhs i don't know where i'm using these examples ai and the nhs sure and if it's you could sandbox it in other areas but i don't i mean maybe sandbox is the wrong word well it feels like we're just educating some grey hair to be able to make smarter, quicker decisions, which just feels a little bit weighty when we should be focusing elsewhere.

46:07I think it's just much too pedestrian. And unless we up the level of ambition, we're not going to get anywhere. I mean, there are lots of places where coordination in itself can be useful. I mean, take the other one on transport, transportation, autonomous vehicles, where if you need to have autonomous vehicles driving around, well, you need to negotiate with the Department for Transport and local councils and the police and insurers and lots of things need to come together. Isn't Wave coming to London, Mads? Waymo. Waymo, sorry. Waymo. Well, that's what the marketing is telling us, and I'm looking forward to it.

46:38I think it's terrific. But we also want Wave to accelerate, right? Kind of the homegrown tech. And we want to make it easy to experiment and deploy. And could it be useful to have somebody from central government helping to pull all these things together to make it easier for businesses to experiment? I think so. But I also think kind of the important thing here is how can we get some of these technologies more quickly into the field? And yes, that should include both diagnostics and therapeutics. But maybe the word should be used, which is probably a word politically dangerous, is concierge, which is like, which is, I guess.

47:15I can just smell Baroness mode in the background. I'm starting to get back. I know, I know, I know, I know. But actually, there is like something that basically can accelerate these technologies into the government's hands? I think it's to start with having business savvy people in parliament. And then secondarily, maybe to your point, have people who are more technically proficient. But this just feels like a barrier. Today, the Ministry of Justice in the UK just signed a big deal with OpenAI, which basically just looks like an enterprise deal by OpenAI. So good for them. Andrew, tell us some SF stories.

47:52Because obviously, you know, being in the land of sunshine and glory, you're going to you're going to have all of the insights and know everything already in the 12 hours that you've been there. Well, a bunch of imperial founders in deep tech were at EF, EF office yesterday, showing their wares. Decent turnout. But yeah, not a lot to report until I start hitting the pavement today. Yeah, get out there. It's morning for you, isn't it, at the time of recording? Well, anything else on the government initiatives before we move on? Because I do want to talk about the 28th regime before we run out of time.

48:24Any more for any more on this? I'm going to move swiftly on. So let's talk about the 28th regime. Now, for those that don't know what the 28th regime, a quick reminder, it's an EU-wide simplified startup entity designed to help startups operate, sell, raise, cross-border. So it's a bit like a Delaware C Corp. It's a very simplistic way of having one entity, one tax regime, one option scheme, and just a very smart way of operating across Europe if you're a startup. Now, the challenge is, I've been reading this week that there is a rumor that it's going to be a directive, not a regulation. Now, that's not as bad necessarily as it sounds but eu inc is pushing for um a much clearer and more concise definition and they've done all of the hard yards so eu inc now you must look at what's the url it is i think it's proposal.eu-inc.org if you haven't and you can see it's a notion page you can see that the guys over at EU Inc.

49:29have done all of the hard yards with 15 ,000 signatures and many legal volunteers working on this document. It's proposal.eu-inc.org. They've done the hard yards to basically be able to present before January the 15th, when this new EU commission comes together, to get this pushed into a useful format that will really make sense for founders and startups and investors across Europe. So my question for the room is directive versus regulation. What's the difference? Does it matter? Would a directive be quicker because every country can make their own version, even though they are abiding by the essence and the major food groups within it?

50:13Is it just too much of a mess if it's a directive rather than regulation? So Lomax, do you want to kick this off? Yeah, you're reminding me of my like, you know, EU constitutional law, you know modules back in the day look okay so as a reminder a regulation is something which is immediately legally binding and we have all of these member states 27 member states when when actually this gets passed it's immediately binding in all of the countries the directive is the watered down version which basically gives each individual country a timeline which to implement something and they can do it in their own individual way if you're serious about the 28th regime and I you know these guys have moved incredibly fast is this this campaign's really only been going for about 12 months and they've already got it to they're now finished the consultation and now moving actually towards the implementation phase if the EU is serious about this thing then it has to be a regulation I think it's very very clear you know for me I I don't want to um pour cold water on it I sort of feel like the status quo for us are basically investing in US, like in Delaware companies, is basically fine.

51:20Like it's fine. Like it's just one thing I would say. The thing for me that's the biggest problem is the ESOPs, sorry, is the employee stock option schemes, which are all different and they all have different tax rules. And the problem is the 28th regime cannot fix the tax because each member state has their own rules on tax. That's not a kind of EU that's delegated to each EU member state. So that's a longer burn. I'm personally more excited about a time when Europe gets its act together on the taxation of stock option schemes. It will be great to have the single company legal entity thing. That is all baked in.

52:02No, no, no, it's not because I said it's not. It's baked in on their website. But I don't think it's going to go. It's not going to be part of the first wave of changes because, as I said, it involves tax, which is absolutely an individual member state and member state thing. But with your legal head, directives aren't as bad as the... This is my impression. Tell me if I'm right or wrong. They're not as bad as you might think as per... If it's regulation, it's just law. It's just how we do things around here. But directives, I got the impression that you still had to follow that. The directive was still available.

52:34You still followed all of the salient endpoints, almost the law, if you like, but you just did it in your own way that served your country, however. but the but the major listen to what listen to what you're saying dan you're talking about giving rain free reign to 27 member states to interpret the rules and implement it on their own timeline that's never going to happen or it's going to happen in a very water down right you know i guess i guess the i guess my kind of feeling behind the feeling was if it's a directive to start with a it'll be quicker b there will be something that is legally binding that's out there but then i don't know what mess ensues after the fact look i'm i'm sure it's not the end of the world this is all good direction but i'm afraid no it needs to be it needs to be in in european law but as i said i'm more excited about stock options i'm less actually i don't really i personally care less about this legal entity thing no i well i'm not with you on that one because i want to see i want to see europe i'm fine investing in denner where c-cops i actually totally fine with it it works fine it's quick simple cheap europe baby europe give it give me more europe mads mads what would you focus on?

53:38Yeah, I mean, I think it's a super interesting discussion. And I think it's worth pointing out that yes, regulation is important here. But even that isn't magic. GDPR is the kind of the flagship regulation we've seen in the last few years. And even with that uniform GDPR, you still had 70 plus opening clauses with national kind of specifications. Germany ended up implementing more than 200 laws to tailor GDPR at a federal and local level. Because the thing is, every new law has to fit into existing law. And so you need the scaffolding. Let's say, for example, some of the financial regulation. But we've seen the kind of things like MIFIT being a very esoteric financial regulation about how the financial markets should work.

54:28Well, it turns out that once you pass laws around that, you need the local regulators to work with them. And they need the laws, they need the legislation to then figure out how to implement and interpret that and work in the framework of all the other laws you have. So this stuff is fiendishly complex. And I think there's a huge risk that even if the EU Inc. guys get their way and get this to regulation status, you're still going to see lots of local tailoring. It's really a can of worms. So, you know, all kudos to them for trying. I'd say, what is the prize here? I'm not sure this is really about a European Delaware as such.

55:07I think it's probably going to be more piecemeal. But even so, if this could mean upgrading the lacquards, if this could mean transforming the German corporate law to say, yes, kind of the GMBH, kind of the traditional incorporation model in Germany, the notaries, the issues you have with stock options, all that stuff, will upgrade that to a new EU model so you can choose between the two, that could unlock massive productivity gains for Germany and therefore for Europe. The European Union, the commission is assessing that more than 2.1 billion euros a year are wasted just by admin stuff, which is a massive chunk of change when you think about it for early stage startups.

55:49I mean, think about if we had another 2 billion euros for seed stage companies in Europe, that would be meaningful. And they're assessing that 800 million of that is in Germany alone. So if this does nothing else than just fix Germany, that in itself would be huge. It's actually a very, very good point. Because if you go lowest common denominator, let's just fix this across Slovenia, Slovakia, everywhere. No disrespect to those countries. But if you just did Germany, France and Sweden, for example, bearing in mind you think the UK is already fine, then you've been moving the needle massively. I mean, getting rid of the need to stamp a wax seal and, you know, have your dog sign it with his paw in Germany and then go and get the local vicar to, you know, bless it would be fantastic.

56:38But I think... Are we still talking about Star Tuts? Andreas has done an amazing job with this initiative. I mean, this is something which if it was executed on in the right way, so it was everything you said at the top of this segment, Dan, and it was simple and it was like a Delaware Seacore, then it would be transformative for Europe. But I doubt it's going to happen. I think we're seeing the first signs that it's going to be watered down. Just enough control will be put back in the hands of nation governments that they fuck it up. And so that it's no longer simple and it will be a waste of opportunity.

57:17I really, really hope I'm wrong for Europe because I don't think people understand how foundationally helpful this would be for the ecosystem and for businesses in general. but it requires an awful lot of change and an awful lot of tweaks and for each individual country to align. And I think it's unlikely. I am not going to end on that note. I'm going to end on a more positive note. What I would say is if you want to end on a positive note is what the EU Inc guys have done is an amazing thing, actually. So kudos to them. And if this can help unlock Germany, that would be huge for Europe. thank you gentlemen for that for that upside upside little upside and a negative and a negative of the uk because uh currently there's still great benefits to being a uk limited company you know that's basically your second choice right if you're not going to be delaware us then your best choice is uk limited so maybe the politicians will eventually wake up and realize what the rest of us did a long time ago which we should rejoin the sooner the better oh my goodness that's a whole podcast in itself right lomax you've got you've got a deal of the week for us before we tail off well deal of the week and actually lots to lots to pick from because i think revolu is supposed to be raising money at 75 billion wave is supposed to be raising a massive 2 billion having raised i think they raised a billion last year didn't they so wave is on doing great things and only enter the japanese market six months ago and i was already watching a video of of the founder and masa navigating the the narrow streets in tokyo so good for them but deal of the week has to be co-mined which is a i guess you call it a health tech company i think they they um i think they have aspirations more in the kind of bci like brain computer interface space i guess you could call it a sort of non-invasive neural link for europe is how you could you could you could pitch it so they just raised 102 million dollars led by plural well how would you describe plural i think they described themselves as the founders fund for europe but they just read raised 100 100 million dollar series a i guess a company founded in 2018 and co-mind at the moment is very focused on health right so a bit like neural link so they have a non they have a non-invasive brain monitoring device that uses low power infrared light and optical sensors to measure cerebral blood flow intracranial pressure and auto regulation this will be a medical device it needs to go through fda in the us that's what the use of proceeds are these clinical trials are very expensive i think co-mind has aspirations for a sort of broader non-medical use case on on bci but um we'll see but that's a big raise nice so uk-based hot tech loving it co-mind go for it gents traveling will breeze look after yourselves get some sleep this has been upside we'll catch you next week all right next week

1:00:19I'm sorry!