330. Meet the YC-Backed AI Founder Who Turned Down A Cambridge PhD w/ James He | Artificial Societies

0:00I'm James. I'm one of the founders of a company called Artificial Societies. What we do is essentially we build large networks of AI personas, digital twins, to simulate a real world audience. And we help enterprises test how their biggest decisions would land with the groups of people that they care about before actually going out there.

0:22Amardeep Parmar:And we were talking before about how you've experimented a lot in the last couple of years to get to where you are today. Can you tell us about that process of experimentation and where you started? Oh, yeah. Artificial society is very much started from research interest. So I was a computational social scientist, which means I studied how do we use machines to predict humans. Nowadays, it seems very natural to try to do that with AIs and everything. But back in 2019, it was not that obvious. I was quite interested in that. And back in Cambridge, I did a lot of work on, okay, how do we create algorithms that can predict the spread of information?

1:02That's what led me into building a prototype concept of, hey, what if we can simulate how information spread inside an artificial society before you take it out there? It was an academic concept. I had an instinct that this is something useful. And I decided that the best way to build a new technology is not by chasing citations inside academia, but to actually make it useful and make revenue from it, because that's the most brutal loss function. So the biggest struggle of artificial societies in the beginning was that everyone say, this has so much potential, you can do so much with this but what's the killer use case we were the startup that young vcs learn to not invest right it's a uh it's a hammer looking for a nail never invest in a company that's a hammer looking for a nail and in our case we were not just a hammer looking for a nail we were a hammer on paper looking for a nail so it was really difficult in the beginning because we were just like, okay, this is clearly something here, but what are we doing?

2:15So our approach was, okay, let's take the OpenAI approach. OpenAI had a cool technology, which they called, catchingly, GPT. So they want to find out what is the use case there. So let's just package this into something that anyone can use, a horizontal product. And then we can see what is the most powerful use case from there. So that was sort of our strategy back in YC. It was building different demo products that showcase what is a simulation of a society, what is a simulation of a group of people and their opinion, and see how the market reacts. We experimented so many of these different products.

2:57We've learned a lot of what to do and what not to do in that. And then eventually, at the start of this year, we just immediately felt this market pull that the most consequential decisions that people had to make for their business, for their organization, and sometimes for their country are the kinds of decisions that require having a societal world model. and the core learning is that for the core learning that we had from building all sorts of experimental products you get to today is that if you are trying to build a new technology you have to look for the use cases that demand this sophistication you can't build a new technology based on something that is unsophisticated to solve an unsophisticated use case and that's how we got you today keep finding the most sophisticated use case that gets the most value from your product because that's the thing that's going to drive you further one of the hard things

3:57Amardeep Parmar:as well i guess is like you when you're building all these different products yes knowing when to let it go right oh yes when if there's a bit of demand for it but not enough that must be the hardest bit how do you how are you kind of going by that process there of those all this could be it but i'm not sure do you just keep it going too long or do you are you quite good at cutting off if you're not too sure it's probably because it's uh if you're not too sure it's probably not the direction that you should go i think for us to know how we to know when to let go of a experimental product is basically what's your north start what are you building this product to do and sometimes you know we've killed products that are profit generating on a unit basis very profit generating, but it wasn't pushing our technology further.

4:48We know that we want to build a societal world model. So what are the products and what are the revenue that can lead us to building this thing that we want to build? Sadly, that's not simulating social media posts or simulating how a Instagram ad will land. That's not going to push us further to building a societal world model. It's lucrative. It's useful. A lot of VCs will scream at me right now, but that doesn't push us to build a technology. And I have to make the opinionated take of, fuck that. I don't care about those revenue. I want you to look for the revenue that push us to build a more sophisticated product.

5:30Amardeep Parmar:And guess what? You did these products as well. The revenue said that. Were some of these products then making revenue and then you cut those streams off? and how i guess it must be quite confusing when you're trying to build something and it's like okay yeah we're making some money here yeah we're gonna stop making money there yeah to do this and like how did that kind of process yeah let me try to think back to i mean the biggest one was that you know we we cut we we spent a lot of money developing and launching a flagship product in the middle of 2025 and it got a lot of traction it got uh it was generating a decent amount of revenue.

6:06And we were frankly approaching profitability at that point along that product line. And we killed it all. At the start of 2026, we were just like, this is clearly not growing in the same level of growth that we think our technology should be growing at. And it's not driving our technology further. Okay, we just have to make the decision. It was a tough one. I don't know how. Let me try to think about this because you're right that it is a tough call to make. You are killing an entire product. You have so many customers using it. And I think we actually just went for a walk. So we were in SF actually because London winter is not survivable.

6:55So the company was still lean enough. The headcount was still small enough back then to just take everyone to SF. for a couple of months to escape London winter. That was really fun. And we were there, we went for a walk. We were just like, do we think we have PMF? The answer is no. Are we excited about keep building this product? The answer is no. Okay, if we got fired by our board and someone else got brought in, would they keep this product? Another way of asking this question, if we were to start this company again today, will we build this product? The answer is no. All right. Then the question is, how will we kill it in a graceful and forward-looking manner?

7:47And that was the decision.

7:49Amardeep Parmar:And tell me about co-founders as well, because obviously, they made that decision. They've got a really good relationship with their co-founders to make it all work smoothly, right? Yeah. I'm really lucky. I'm really lucky to have my co-founders. So we didn't meet in the traditional like co-founder matching programs or whatever. I actually met my first co-founder, Patrick, because I was trying to sell to him. So it was like janky prototype and he was really sold. So he already knows the industry. So he's very much coming from the traditional world into the startup world. I think in the beginning, we had to navigate a lot of that because he's used to doing things in the traditional way.

8:32And I've been lucky to be in startups before. So there was a lot of navigation. And then our third co-founder, Tom, first joined as a founding engineer, actually. And after YC, we were just like, Tom poured so much of his life and his soul into this product and this company. We can't imagine building this company without him. So let's make him a co-founder. And then people told us that that's a really rare thing to do, to make someone your co-founder to recognize their contribution. And I was like, well, do you want them to keep contributing? Yeah. And then Tom is now doing phenomenally because he's now, so while Patrick is spending more time deployed to clients, Tom is taking up product along with tech, along with design, along with everything else.

9:21It's amazing. so I think the three of us our superpower is how much we listen to each other how much we communicate to each other I think a lot of founders forget just how emotional the journey is because you're talking about something that all of you care deeply about and sometimes the same level of care manifests in different ways. So when I say I really care about artificial societies, it manifests in working 24-7 non-stop. When Patrick says he really cares about artificial societies, that it also means working hard, but it also means that he's taking time to specifically maintain his relationship so that he can focus better at work.

10:10At first, I didn't really clock that so there were a lot of like friction of like patrick why are you taking holiday again and it's like um well to be a high performing athlete recovery is just as important as training and i think that's just something that through communication that we learned there are so many moments in building this company that i'm just like wow like i'm so grateful for my co-founders and for my wider team, I don't think we will be here without them.

10:42Amardeep Parmar:And you said, well, so in December last year, when you did the trip to SF, the team count was still quite low. Yes. But then this year, you're now expanding that, right? Yes. So tell us about that, how you've grown that team and how you think about that strategy as well. We think of capital as ammunition. So we raised just over$5 million at the end of our YC badge back in March 2025. but then we stayed to like a core team of four or five people essentially until march this year so for a whole year we stayed very very lean because um in our philosophy capital is ammunition preserve ammunition until when you know where to spend it and then you spend it so for that year for that 12 months we were doing different experimental products we were still finding out what is the best deployment of this technology.

11:38And I think today we have some early answers to this. We are now informing things like, how should you restructure a company? How should you communicate to the US Senate? And how do you navigate things like the Strait of Wormuz? The kinds of decision that makes a needle, that moves the needle of the global GDP. so we are lucky to have found this now so now it's like well this is the time to back ourselves let's deploy some capital and look we have a lot of those that we can now deploy so now um building the team hiring is has been a big focus for myself over the last few months and what are you looking for in these team members spikiness spikiness is a big one if you give more concrete examples.

12:26Our company is called Artificial Societies. And in the name, you can already sense that it combines this idea of technology, there's a techie side, and then there is a humanity side. We are all behavioral scientists and market researchers and social scientists, political scientists by training. So actually, no one in the core team right now has a computer science degree. We are all social scientists first, and then we learn the skills to help us unfold humanness better. So when we're looking for the new team member, a big part that we're looking for is that combination as well, is that interest in understanding humans, and also the steps in understanding technology.

13:10That's proving to be probably the biggest challenge because not many people have that intersection. Or, you know, there's a lot of startup folks who's read a few books on popular psychology and they think they know everything about cognition and human evolution. I'm like, yeah, there's a lot more here. You have to have actually studied human behavior to actually know what we are talking about. So that's one thing that I'm looking for. I think another thing I'm looking for is just high agency. Like we are inventing something that is fundamentally very new. You can't be someone who's lagging behind you have to really push for what is the new boundaries.

13:51How do I apply this technology in the most creative way possible? There are things that we have never imagined before, because a lot of the time we look at what we build and say, oh, this is a much better, this can replace a lot of human panels. But no, no, no, no, no. That's just thinking of what we build as a faster horse, right? We are not building a faster horse. We're building a different mode of transportation. We're building a different way of getting insight for the decisions that matter. What are the other ways that we can use this now that you can run infinite experiments inside an artificial society of your customers before you actually experiment with your actual customers?

14:35What can we do with this? There are creative ways of doing that that requires spikiness that requires creativity and agency. Hello, hello.

14:46Amardeep Parmar:I hope you've enjoyed the show so far. I'm Amadeek Parma, co-founder of BayHQ and the host of this show. For those of you who don't know, BayHQ is the community for high growth Asian heritage founders and investors in the UK. Over 7 ,500 people have attended our events, 200 people have been for our impact programs, and obviously there's been over 250 episodes of this podcast. If you want to join us and take part in the programs and come to the events, go to behihq.com forward slash join. We also just released our first ever book called Startups for Outsiders, which you can get on Amazon now and the link is in the bio.

15:26Amardeep Parmar:Hope to see you soon at our event. Hope you enjoy the rest of the episode. And on that side too, you said you started off as a behavioral psychologist, you've done research and all of this. But now being the position you are, obviously you're going to have to update so many of your own skills, right? And what part of that journey and your personal development do you think you've improved a lot on now compared to when you first started the business? Where do you think that you had to make the most improvements? I think communication is definitely the one that I had to make the most improvement on personally.

15:55If you talk to my old boss at the previous places that I've worked at, they will say, yeah, James needs to know that to lead a team, it requires you setting goals and be more pragmatic. I think, yeah, so communications and being more pragmatic are the two things that I've improved a lot upon. To be more specific, when a team member isn't understanding you, there is a reason. It's probably because you didn't set the context right. So, okay, let's set the context here. So we want to enable this customer to be able to run simulations on their own because we want to get to that point. And then they understand, ah, this is why I'm being really pedantic about that technical specifications.

16:44And then the other learning is pragmatism. I think myself as an academic back in the days, academics like to do things, do fancy things. So we like to maximize sophistication. and even my old job, I think I fell in the trap of maximizing sophistication. Oh my god, wouldn't it be cool to do this? There is a value in having sophisticated things. That's what got us to having such a sophisticated technology like artificial societies. But at the same time, you need to be pragmatic. Are you actually building something that is useful for your customers? So one of the most brutal learning that we had was that you need to use revenue as a judge of whether you're building something useful.

17:27So only build something if you can forward sell it. So instead of, okay, I'm just going to create this crazy product and I think the world is going to love it because Steve Jobs, when I've done this, well, you're not Steve Jobs. So many founders just don't realize that, right? It's much more dangerous for someone who's not Steve Jobs and think they are than for a Steve Jobs-like founder to think they're not, right? If you think about the impact to your business. So like for us, we realize that we are probably not Steve Jobs. We try to be. We try to give our opinionated take of the market. But we also just listen to the market.

18:14Like, okay, let's say we build this. We'll just go to a customer and say, we are, you know, this feature is going to become available in the coming months, what else do you get your take? If they're like, oh, can we have this earlier than you release it to other people? Hold your horses. Yes, that would be$10 ,000 extra, please. And they go, okay, cool. I'll think about it. It means that no, they don't want it. If they say, right,$10 ,000 is too much. Can you do$2 ,000? that's actual purchasing intent right then you actually measure okay that okay building that feature moves the needle by two thousand dollars great then you know the magnitude can be very different sometimes it can be two million it can be ten million and uh you just have to be very pragmatic about are you building something that is generating revenue not because you are a greedy bastard but because revenue is the only filter for what matters to your customers if your customers are not going to pay you, it means that it doesn't matter enough to them.

19:20So you just have to focus your energy on building something that matters. And it's the hardest thing with that is that people say nice things, right?

19:26Amardeep Parmar:Oh, yeah. But they didn't mean it. If they're not going to pay for it, they didn't mean it, right? Exactly. You know, we love our customers. They give us, because they give us so many very, very constructive feedback. And yeah, you don't want people to just say nice things to you so that's why pricing pricing is such an art prices are things that filters for what matters so gotta price this stuff and you're now talking to the enterprise clients right how did you start like those initial relationships and start to because obviously the famous thing is that enterprise clients take like six months nine months even longer than that to get through yeah how did you go about that how did you kind of crack into some of these companies i think we're living in a different world now i don't think enterprise takes six months to close now.

20:10I think enterprises are in fact under a lot of pressure to try new things, to experiment. And so if you're building something new, I bet you there's a lot of enterprises who are really interested in what you're building. And I bet you maybe you'll have a friend who knows someone, who knows someone who can get you that introduction. And I think a lot of the time, you know, people think about enterprise sales as very much like you do cold outbound or you do a warming tro and you build relationship you do conferences sometimes consumer channels can also help with enterprise stuff because it's just about awareness building you're building something so new that people are not aware of you so by having more consumer touch points means that you will earn the right to be in front of enterprise customers as well so for example right now we have an ignitable outbound effort all inbound and given the all inbound i am trying to scale the team to meet the demand so you can you can imagine like like the market is very much uh ready for new things and they hear from each other so much you said about building that awareness piece how did you build that awareness before you had customers i think we were really lucky that YC was good.

21:27Building in public is a good idea. I'm very lucky to have a decent LinkedIn presence, a decent online presence in general. Just tell your story because eventually humans buy from humans. Machines maybe buy from machines, but humans buy from humans. If you think about why does someone choose a product, it's not just about the functionality of that product. It's about the personality. Eventually, they want to buy this, you know, I want to buy an iPhone because of the personality and the opinionated choices that the designers of the product, well, eventually Steve and his ghost made, right? So it's about the human side.

22:14So if you tell your human story, then it attracts that human resonance that's what's worked for us i don't know if this is the most optimal one in fact this morning we were just doing a workshop on how do we rethink our go-to-market strategy so take it with a grain of salt or a pint of beer but yeah i think my instinct

22:39Amardeep Parmar:is that humans matter and with that storytelling right so with linkedin or wherever you're posting as well what is it that people seem to be resonating about like how is that then converting what is that about your story that you think actually people really relate to i think there's two parts perhaps why what you'll probably have to ask my readers i don't actually know uh well i think the one part is the founder story is something that a lot of people aspire to and a lot of people find resonating and the other part is the builder story where you are building a new product, you're talking to customers, and you're discovering new things.

23:15And sometimes it's, you know, none of this is actually talking about a feature or something like that. But you build a brand. And then once or twice, when you drop a video about, oh, like, here's us doing this thing to solve this actual business problem, your followers will go, oh, actually, well, that's really useful. Can you do that for us as well? Can we chat about this? So I think that's how we've been getting a lot of inbounds but as you can tell i need to do more homework on how exactly

23:45Amardeep Parmar:how do people find us and also so you lived in sf for a while with yc you were in sf over winter there's this whole kind of linkedin thing going on at the moment about how everyone needs to move to sf and everything's back to there but you also have the counter thing about london maxing right yeah and how do you think about that like what makes you based in london most of the time and why is this the best place for you to build right now well we have global customers just from a very practical point of view. When I'm in SF, I get up at 6 a.m. in the morning and I'm already being attacked by 70 emails.

24:17I am not ready for the day. In London, I can just fend off the Asian and Middle Eastern customer and I can attack the U.S. customer just from a time zone point of view. When I'm on the West Coast, that's just like, just in a very pragmatic way, it's really difficult. I think just on a broader point, London has a great intersection of all these different industries. And I think the London community is getting better as well. There is something to be said about, so you should pick the place to build your company that gives you the highest chance of success. That's what YC told us. And then the next sentence that they said was, most companies succeed in SF.

25:04well the counterfactual here anyways they didn't do the stats the right way they took that up a little bit doesn't sound like storytelling storytelling stay close to your customers stay close to your talent and when I look at my customers spread out around the world okay being in SF means that I am one day late to everything that they send me that's not great and talent. SF has a lot of talent, but we want computational social scientists. Some of the best computational social scientists are in London, are in Cambridge, are in Oxford. And we want to have a company that isn't detached from the real world.

25:53You've been to SF. that's not the real world. People there live in their own bubble. They have their own fears that is amplified upon their own assumptions. They don't know what a grass is. I think it's important for a company that is trying to model human opinions and the texture of human nature to be close to the texture of human life so i think london is the best place

26:26Amardeep Parmar:for that and you said as well about the global customers right yeah so was that an intentional thing to have like customers all across the world or did it just happen because that's i think it just happened um again we didn't have an album strategy which is terrifying so i'm now forming one. But yeah, no, so it just so happened that there is demand around the world. I think what it tells us is that what we are building has demand across different markets, across different sectors, and across different teams. And again, terrifying because then you need to narrow in on your ICP, right? So what is your ICP?

27:03So we do a lot of work around essentially what type of decision maker and how are they making decisions. That's how we define our ICP right now. so um but it turns out that that's still very global it's just the natural result and obviously