Marketplaces & Network Effects In The AI Era With James Currier From NFX | EM Group Chat #208

Welcome back to Everything Marketplaces, where we talk about founders and leaders on some of today's top marketplaces. So this is episode 208, which is really a group chat to have with James Currier, who's a founding partner at NFX. NFX is a venture fund that's backed marketplaces like DoorDash, Lyft, Poshmark, Outdoorsy, and more. James is the top mind when it comes to both marketplaces and network effects. James is also a past group chat guest. This is great to welcome back on, where we started off with a quick overview of NFX as a venture fund and what they look for in marketplaces when investing. We did a deep dive into marketplaces in the new AI era. We discussed topics like network effects, defensibility, the current state of fundraising. James shared lots of tips for founders, and we also had a great group Q&A. I did want to mention that we ran into issues with the Zoom and had to make a few edits, but I think you'll enjoy this conversation and find it a great watch to the end. So, James, it's great to have you join us again for the follow-up group chat today. And, you know, I want to start off by saying huge thanks for taking the time to do so in advance. You know, I feel like so much has also changed since your past chat. That was, you know, 2 years ago. We of course have a lot that we're going to dive into with this, you know, with marketplaces and network effects in the new AI era. But I thought it'd be great if you could start off with a quick refresher on your background for those that have just joined us here.

Sure. So my name's James Currier and I'm the co-founder of NFX Capital, which is one of the largest seed funds in the world. We have been sort of on the cutting edge of marketplaces for 20 years in terms of angel investing. And then once we started the firm about 10 years ago, about For a while, about over 50% of the companies we invested in were marketplaces and we've published extensively on nfx.com. I think if you look at all the VC websites in the world, Andreessen is number 1 in traffic and we're number 2, and then Sequoia and then First Round. Typically the 4 of us are at the top every time. So there's a lot of content that we write about marketplaces and AI and all those things. And so people send that content around a lot and hopefully help make them, you know, create better companies so we can invest in them. And prior to that, I was the founder of 4 venture-backed companies that all did very well and allowed me to, to do this later in my career. I got off the field and now I became a coach.

That's a really great background. Of course, you know, NFX and all, all your different resources and, you know, posts are commonly being shared in the, in the community. And, you know, you briefly mentioned, you know, an overview of NFX as a venture fund, but, you know, how do marketplaces fit into the investment strategy in 2026 with that AI?

So, you know, we, NFX stands for Network Effects and Marketplace is Two-Sided Marketplace. network effects, 3-sided marketplace network effects. These are, you know, 2 or 3 of the 17 network effects that we've looked at and identified and chronicled. And so we're always trying to find marketplaces. Now, that being said, the percentage of companies we've invested in that are marketplaces has been dropping. We've noticed that over the last 5, 6 years, not because we want to, just because as we look at each company and decide, can this really be a business? where we're realizing now they can't discount it. So we're kind of unfortunately doing fewer marketplaces because marketplaces are such a great network effect and it's where you do get your multi-billion dollar outcomes, which is what we need as a venture firm to actually be able to make our business work. So we keep looking for marketplaces and we think that AI should enable all sorts of new marketplaces that weren't possible before, primarily because AI can read Not because AI can write, but we can get into that.

And we can keep this a little bit more, you know, high level because I know you also shared your marketplace scorecard in the past and, you know, we discussed in a little bit of detail on your past group chat. But, you know, what do you look for in marketplaces when investing?

Oh, there's so many things. Unfortunately, because we know them so well, we can find the flaws in most of them. You know, you know, a lot of the, a lot of the marketplaces don't end up working, particularly on the B2B side. Because every node in the network wants the exact same thing, and that's margin. The consumer marketplaces tend to be a little easier because the consumers want variety and they want exactly what they want. So it's basically selection and value, whereas the seller wants a high price. And the, you know, so the consumer's willing to pay an extra $8 for the shoe they want. $12 for the shoe they want, and that provides the margin for the seller. On B2B, we have this challenge where everyone's going after the margin. That's what they're all there for. And so B2B marketplaces have proven a lot more difficult. And, and that's why we have, we have, we have persistent brokerage systems in many of the regions where you might expect for B2B marketplaces to form. But in general, another reason that, that that marketplaces fail is because one side or the other is too concentrated. You know, you'll end up with 5 buyers of data for Scale AI and Mercore. And at some point they're going to stop buying and those companies will implode. And so we need to, to understand marketplaces in that regard as well.

Yeah. And I wanted to transition to, you know, talking about marketplaces in the new AI era that we're in. And, you know, things of course, you know, change so quickly since your past chat and, you know, it's transforming marketplaces. As far as, you know, AI. So, you know, where are we at with marketplaces right now?

Where we're at is, you know, if I look at my last 7 investments, I think 4 of them were marketplaces. But the overall— the firm's investment in the marketplace is going down as we're into more bio or into more space or into more deep tech that don't have marketplaces. We're into more sort of, you know, AI infrastructure tooling. And so I think the opportunity is there. I think you just have to be very clever about how you use AI. to enable a new type of marketplace that was impossible or too costly in the past. And that, you know, we're looking for those types of businesses.

Yeah. So I want to expand on that a little bit more. So like this, like compelling, like why now for marketplaces? Are there a few examples of that maybe you could share and, you know, how AI is going to, you know, enable them to be better marketplaces?

Well, for instance, you know, we've got a buy now, pay later marketplace between lenders and consumers in Brazil right now. where they're providing software to retailers and then using AI to make the matches between them. It's like a Lending Club, but more streamlined. And they're gearing up toward becoming the next Visa or Mastercard inside of Brazil because AI can look at so much more data on the consumer to get a better, accurate picture of their credit risk and then can figure out what fuzzy logic the lender is using in terms of where they want to use the where they want to make their loans. And then matching those at scale, at speed is something that was really untenable. It was imagined about 5 or 6 years ago and they were kind of cobbling it together, but now it's definitely doable. We've got another company called Rec.us, which is doing software for parks and recreations departments, and they're building a marketplace on top of that. Once they control the real estate asset, once they control the facilities with the SaaS, They can then build a marketplace on top of that for coaches and for players and teams and leagues and, and whatnot. And then the pricing moves and then the timing moves. And AI is helpful again for that fuzzy logic understanding and also for onboarding a lot of consumers, onboarding a lot of coaches, ingesting a lot of data about the facilities. The AI again can read really well. And so the friction for this to occur has really dropped. And so we think there should be an explosive recreation marketplace in the United States over the next 5 years.

It's cool to see, you know, what, what's possible now with AI. So, and you know, as this starts to like kind of transform marketplaces, you know, how, how do you think, like, I would say like the value that they add is going to kind of start to change and like ultimately, you know, the role marketplaces play?

I don't think it changes. I think marketplaces need to stay the place for people to get smooth liquidity, right? The easiest, Liquidity, lowest price, highest price, most volume. All those features that marketplaces have created since we had the Agora 4,000 years ago, the same value will accrue to marketplaces in the future. I think it's just AI changes the speed and the accuracy and the low cost to get into a marketplace if you're a new node on the system.

And I would say one of the things that we're really seeing in the community and often discussing is the barriers being So just easy, never to kind of, you know, build a launch and also to kind of sense of speed. So, you know, like with this, I would say like a lot of these kind of new AI native marketplaces are not taking that kind of more traditional approach as well. You know, going from 0 to 1. So like what are some of these like early strategies maybe that you're seeing like marketplaces kind of take and that seem to be kind of like, you know, playing out right now?

You know, look, I think the, if you go back to the Nanofex article, 19 Ways to Solve the Chicken or Egg Problem, you know, the AI can affect about half of those. So the fake it until you make it type of thing where you use AI to create a database which looks like the supply is on your marketplace, that's a lot easier than it used to be. And we're seeing that, we're seeing that play out repeatedly. The other thing that AI is doing is it's better at re-engaging supply. So you basically, or demand, so you extend the amount of time that a person will be engaged because you can write them a better email, you can write them a better text message to re— bring them back when an item they want might be there. And so one of the challenges with a lot of marketplaces is timing, meaning you've got to get enough people on the buy and the sell side at the same time. But with AI, you can now extend that from the same 3 days that they're engaged to maybe more like 3 months. And so it, it makes it easier to, to get that liquidity kicked off.

Definitely. We'll have to, we'll have to reference and kind of revisit that, you know, the past articles as, as you mentioned. And then another topic that I wanted to really spend a lot of time discussing because I feel like you're probably one of the top people to discuss this with is network effects, right? And so how are you kind of seeing network effects change in this kind of like new AI era?

So I don't see that the network effects themselves are changing. I just think it's getting harder to make those defensibilities work because in the past, the only nodes available to you were humans. Now you can have agents as nodes on the network and another marketplace could generate 100,000 agents as well. So if you have a marketplace that includes agents, it's now easier for competitors to get to critical mass, get to that threshold to then compete with your marketplace. And so we are puzzling through with each individual company we look at whether they can get to the threshold and then whether they can maintain it with any reasonableness. Given that AI makes it easier to jump into a market and compete. And it, you know, so that's just something we're evaluating. We're just looking at curves and numbers and data to figure out where that line is. And we're not going to be right all the time, but maybe we'll make an investment and we'll be wrong, or we won't make an investment and we'll be wrong. So we'll see.

Yeah. And I guess like related to this, you know, do you have any tips or maybe advice, you know, for founders to like kind of like take this into consideration? You know, at the very initial stages, like right now when they're starting a marketplace and, you know, designing, you know, designing this with kind of network effects in mind?

Nothing, nothing really new. I think, again, just go back to nfx.com and click down to content and then click into marketplaces and all the articles there, I think, still apply. Again, I think, I think the AI hasn't changed the fundamentals. It's just changed the costs and the speed.

Definitely interesting. So lots of founders kind of taking on these, these new challenges here. Another kind of related topic to this is around the defensibility, right? And I feel like Especially if, you know, anyone's on social today, they're probably hearing a lot about this, right? With AI, you know, so how do you think about this? I would say for marketplaces today, defensibility is the same thing.

You've just got to get enough liquidity so that it's not worth it to go anywhere else. It's that simple. It's about liquidity will give you your retention, liquidity will give you your accurate pricing, liquidity will give you enough data to match. So it's always the same idea. You know, this, this Wrapped.US company's actually making a 3-sided marketplace, which is very difficult to do. We've seen very few companies pull that off because the complexity goes way, way up, but they seem to be doing it. So I think AI will make it easier to go after 3-sided marketplaces.

And, you know, another topic that we'll kind of transition into and hear, is on the note about fundraising. And, you know, I think a lot of founders right now, of course, you know, are seeing or feeling like, you know, marketplaces are in large part kind of like being overlooked right now. And there's this kind of like tale of two cities with, you know, like AI and then kind of everything else, you know? So how are you seeing this impact marketplaces, you know, when it comes to fundraising and at the seed stages?

I think that's exactly right. I think anything that sort of looks like Merckor or Cursor, that's getting all the VC attention. And so if you're gonna have a little something a little more complicated or something that's gonna be slower growing, you're gonna get less love for sure. What my suggestion is, just make sure that in this day and age, like your marketplace is truly being able, enabled by AI and then turn it into an AI story. And, and in fact, it's, you should be able to go faster if you really use AI. So I would encourage the real examination of, of how you can use AI in your market. And then I would also say that, you know, one of the things we're seeing at Series B is that everyone's realizing that there's not a lot of defensibility for Cursor, right? Because their market share is getting killed right now by Anthropic and OpenAI's market share is getting killed by Gemini. And so there's not a lot of defensibility without network effects. And so a lot of the Series B guys are actually asking the question, at what point does your tool get you to be a marketplace. And the, I've had some founders in my portfolio who weren't going to give that pitch in their Series B, but the investors are demanding it because they're now looking at long-term defensibility in order to get their returns. So, so understanding the marketplace's models and how you would build those products and how you would maintain liquidity might actually be a lot of help for Series A and B when at the beginning, It'll be a little harder, but in the end it'll pay off for you.

And, you know, related to this, you know, how do you think AI might start to change the capital needs and requirements for, you know, for startups and marketplaces more specifically?

You know, there's going to be a lot less need for venture capitalists like me in the early stages. If you can get a, if you can afford a Claude code license and you can build the system, and if you have some proprietary access to one side or both sides of, of your marketplace, you can just get going. And then you can raise money from us at a later time at a higher valuation if you even want to raise money. You know, you, you have to get some good viral effects going for your CAC on both sides, which is always hard. But if you are a good viral engineer and good at, good at managing that, then there's no reason why you need to raise venture capital. Cool.

So I promised founders we'd save some time for the group Q&A today. So I want to make sure, you know, that we can do that. But, you know, right before we jump into the group Q&A, you know, what would be your top tips for founders starting AI-native marketplaces today?

You know, look, these marketplaces typically have taken about, 5 years to get going. They've been very, it's been a long-term love affair that you have to have with your market. I think today is a little bit different, but it may be 3 years instead of 5 years. You still have to be dedicated to it and build the nodes and figure out who pays whom and figure out the language. It takes a long time to find a starting point and then a place where you can get and maintain liquidity and then grow from there. It's not an easy task. It's just that the payoff is really big when you do figure it out. And so maintaining your, your understanding that it's going to take a while, I think, is sort of the number one thing. And then the other thing is, is really look to see if there is a set of nodes that are non-participants today. That you can bring in. So a lot of people say, oh, I'm gonna do a, a B2B steel marketplace between buyers and sellers of steel, and they're just gonna use my software. This never works. What you wanna do is find out lumber people who aren't in the steel business, and if they use your software, they use your marketplace, suddenly they're making money where they were making zero before. So find non-utilizers to get newly into business. They'll be a mo— a lot more loving of you. and a lot less skeptical of the margin you take because without you they have zero as opposed to someone who already has something. And now they're looking at the fine-tuning and whether you're gonna betray them and ruin their existing business. And then they have a lot of fear of loss and therefore they resist subtly and not, not so subtly. And these marketplaces take a really long time because of it. So I would find non-utilizers and bring them into a new marketplace that they never knew existed before.

It's gonna be really helpful, a lot of tips. So, and we're gonna go ahead and get into the group Q&A. So if you, if you had a question that was added, you know, feel free to use the Zoom raise hand functionality. That way we can see and call on you to come on. But I did actually want to ask a question that was, that was added in advance to kind of help kick things off. And it was around taking that agent-first approach to starting in marketplaces versus like the more kind of traditional approach. And so when do you think that this works well?

The answer, guys, is I don't know. I think that this whole new agent-first thing is, is brand new and I am a learner just like you are. You know, every 14 years we have a big technological revolution. We had the browser and we got the mobile phone and now we've got AI. I'm learning along with you. This is the funnest I've had in a decade. It was so boring up until 2022, literally. And so enjoy it because this window will shut in the next 2 to 3 years. Like people will have figured out this stuff. I don't know. I am very hopeful about agent-first anything because it's a new paradigm. It's a, you know, it's a, it's a new modality. And I'm just optimistic. I just don't know what the playbooks are yet. I haven't seen them yet.

There certainly is like a, like a real sense of energy, I think, in the community that we can see with founders. And on that note, too, a lot, a lot are kind of taking the agent-first approach right now. So it's super interesting. Hey, Molly, saw you raise your hand. Do you want to come on?

Hi, I'm Molly Morris, the founder of Recess. We're like Booking.com of kids' classes and camps. I'm very curious about the agentic marketplace. So right now I feel like you see a lot of SaaS companies who are building entire companies around some sort of agentic capability. And I'm wondering if there's a future that you see where instead we can sort of rent or employ the agents and embed them into the business ourselves, as opposed to kind of contracting with this entire company. I think the thing that spurred it for me is Boredy. Boredy really wants to like be on your team as a fundraiser. I think that will happen. I mean, you know, I, I can't imagine we're going to continue to have these monolithic software companies, right? That, that we're either going to have what you just described, which is just people, you're going to have an agent which will go out and rent other agents. The problem with the agents that they're renting is that anybody can kind of spin them up. It's not hard to spin them up. So yeah, I definitely think that's happening. And I think Forte is, you know, Forte has pivoted now twice and now they're trying to become a venture firm because they can't figure out what the business model is. It hasn't really taken off yet. So I hear you, but they're struggling with what the business is. I mean, they're, I've run into them at multiple events and they're, they've kind of got a lot of fear in their eyes. So, but, you know, well, I think what they're doing is a very interesting experiment and something to watch like Mulpuck. It's just very, Very interesting experiment to watch. Or we're going to have this world that you're talking about where it's just all these free-floating agents, or we're going to end up having what I call blue and orange software. When you're a startup with one person, you say, I'm a blue software person. And then it watches my emails, watches my calendar, watches my Slack. And then it says, hey, do you want me to build this software for you? And I'm like, yeah, sure. And then it's just going to grow and I'm going to end up with this sort of artistic, unique set of software tools and whatnot. It's just Blue software and I'll just, I'll just, Blue will just grow with me until I'm on 1,000 employees. It'll just keep building more and more software. I won't need to buy Salesforce. I won't need to buy HubSpot. It'll just build me a CRM. So that's, I think these are the two, you know, sort of futures, both of which I'm sure we'll probably see some versions of. But yeah, no, I, I think it's exciting and I think you're right to, to be watching what Porty's doing. I just realized they haven't, You know, we said about 10 years ago, we said we're going to be a media company that monetizes through investments. And Andreessen said the same thing. And now Barty's saying we're going to be an AI company that monetizes through investments. And so at some point it's all going to come down to like equity ownership. I don't know if that's the only way to make money. We'll find out.

Hey Jeff, saw you raise your hand. Do you want to jump on here?

Hi James. I'm an Indianapolis-based entrepreneur building the home improvement space, leveraging a whole lot of AI to make smarter matches with homeowners and their projects and pros. My question's about team. What are you seeing in early teams, like more so at the pre-seed, and I guess pre-seed, seed, whatever, who knows what's what anymore, but what structure are you seeing there? You know, In the last 12 months that's compelling versus maybe what you saw a handful of years ago? A couple things I'm seeing in trends. I'm not sure if this is going to answer your question, but one is people are just getting younger, more aggressive people who just work, you know, 18 hours a day. learning this stuff and are constantly upgrading. And they aren't using titles because somebody has to do this and that and the other thing. They're literally just learning as they go. And it's just a ferociousness about the employees and about the team that is dominating the culture and the way they think about hiring. The second thing, the second trend I'm seeing is Quite successful people who are clinging to titles. And you end up with 8, you know, 10 employees of a company and 8 of them have a chief something title because they're terrified of losing their status in the face of AI. And I'm seeing both of these things. I'm warning my founders against this. Like, people who've been successful in the past now wanted to be called chief creative officer. Chief Revenue Officer, Chief Science Officer, seriously. And as I got 8 people with chief, and because they're all like in their late 30s, early 40s, they're like, no, I'm very successful. But in the face of AI, they're terrified that they're somehow going to slide backwards. And so they're using titles and clinging and trying to adapt, but not actually putting in the work. So those are 2 trends that I'm seeing that I don't know if that's helpful to you, but I would avoid the latter and move with the former.

Hey, Daniel, I know you sent a question in advance. Do you want to jump on here?

I'm Daniel. I'm building an end-of-life marketplace. We help bereaved people find help legally, logistically, emotionally, financially. My question is about ratings and reviews in the age of AI. Have you seen any, like, evolution beyond traditional 5-star system and services marketplace?

Does AI create an opportunity for something else? Maybe More nuanced, contextual, bespoke, or honestly, is change unnecessary? Because stars and numbers still work pretty well.

The answer is I don't have any company to point you to who I think has done a unique job there that you could copy their, their approach. I would do that if I knew it, because I think that's how great product and design is done, is we just copy all the best ideas of all of us, right? It's a community. What I will say is that the The veracity of the postings is now so much more in question because AI is writing so much more stuff. And I am, I am a fan of any system that helps people believe that what they're reading is not AI. The most common thing I hear from consumers in particular Is my life is just AI slop. And they don't use the word slop, but they, they realize that it's just a lot of garbage. And I think it's turning people off of TikTok and off of Reels, and people are just spending less time watching that stuff because it's so filled with garbage. And your reviews can get filled with garbage really quickly if someone has a financial incentive or a status incentive to pump that AI stuff into there. So I think you just have to be really careful and really clear with your interface that this is a real human being, you know, posted on this date. Here's their real name, here's their real location. Like, it's, I think it's gonna get tougher. I think this ratings and reviews system overall will be under siege, let alone the 5-star thing. It's bigger than that. The problem is bigger than 5 stars, I think. You know, you know, the reason that 5 stars works is that most things are pretty good. And so if I get a 4.2 or a 4.6 or a 4.8, the consumer doesn't really care. They just know that it's been vetted, it's probably pretty good and it works out. It's like, it's like your Uber driver. Like, I haven't really had a bad Uber or Lyft driver in years. And I don't know how many bad riders they've had. Most of us are okay. And so I think that system continues to work for the reason that most things are pretty good, but AI is a threat.

James, I actually want to ask a question that, another question that was added in advance and it was around pricing. And they were asking that, have you seen any marketplaces innovating on pricing?

Not as a result of AI. I mean, you know, the thing that The thing that keeps coming up, particularly on the B2B marketplaces, is who pays, right? And then what is the interface that both sides see? Because you want to be totally transparent that this is the price everybody's seeing. But at the same time, if you play with opaqueness there, you can actually start to get margin from each side of the marketplace because both or neither, you know, Sometimes both are willing to pay. So I've seen a bunch of experiments in that that I thought were really interesting. And then the buy now, pay later stuff allows you to increase your conversion rate because you're allowing people to borrow and pay over time. But I don't know if that's really an innovation of the marketplace. It's really an innovation of the buy now, pay later lending systems that exist. And they're getting more and more sophisticated over the last 10 years, which I think has increased conversion in many cases. Other than that, no, the only other thing I would encourage people to think about, you know, Listia years ago and a few others have been playing around with 2 currencies. So in video games, you know, as one video game, developer said to me just last month, he said, you know, I haven't built a single currency video game in like 15 years. Like, of course you have 2 currencies in your, in your game. And, and, you know, the, the last video game I built got to about $120 million of revenue. And, and we had 2, we had gems and gold, and they trade off, they're useful for different things. And so adding in some other sorts of currency alongside of US dollars could be pretty interesting. It's something, something to look at.

Yeah, definitely. That's super interesting. That's an interesting example. So I know a lot of times we're thinking like, what are areas that we can kind of innovate on right now, especially with AI, right? So cool. Hey, David, saw you raise your hand. Do you want to jump on here?

Great discussion. Loved it. So I'm David. I'm a co-founder of Prosperus. We're a credit marketplace. We sit at the intersection of real-time telco and financial data, and we're embedded in apps that have about 100 million users across Mexico and Colombia. I think given the world we live in, the AI-first world, I'd love your take on what moats that you're seeing are proving to be durable and defensible these days compared to what you saw in the past. Yeah, I need to emphasize again that this whole data network effects moat is usually not real. It's very, very weak. Almost every founder I talk to every day says, well, we get more data, therefore we get more accurate, therefore we're going to get ahead. No, you don't. So don't be deceived by thinking, ah, the more people we have, the more data we have, the better our accuracy is going to— not really. Because your matching algorithm is, when you first launch, is already 94% good. And over time you might get 97% good, but your competitor can launch with 94% good. And most people can't tell the difference between 94% and 97% good. So, you know, don't, don't, don't believe that in your own thing. And then, and so in terms of things that are defensible, you know, if you go back to the NFX website, there's a defense, there's a, the title of the article is too long and it's a bad article. It's a defensibility is a founder's best friend or something, or adds defensibility adds the most value to founders, something like that. And it lays out all of the defensibilities and which are basically 4. And then it lays out what the competitive advantages are, which are not defensibilities. There's competitive advantages like raising a ton of capital is a competitive advantage, but it's not a defensibility. And it lays out the difference between those 2. And then go look at the Facebook and the Uber videos and articles on nfx.com. Those lay out for you how you can use the map to sort of Figure out what your competitive advantage or defensibility is, and then augment it and move between different ways of creating defensibility. I think it's very instructive as to how to think about reinforcement and all those things, and which comes, you know, depending on where you start, another one might come next, but it gives you examples of different companies and how they approached all this. I think that's sort of a masterclass on defensibilities. And then if you, if what I will say is that the AI is making it everything less defensible, everything less defensible. And so you just need to be better. It's, you know, it's like playing in the NBA in the 1960s was one thing. Playing in the NBA now is a different thing. Like your skillset just has to be that much better. It's just getting harder and faster and better. And you as an entrepreneur need to be harder, faster, and better. If you end up with contracts with, let's say you've got a marketplace situation in LATAM where 4 banks account for 50% and then the other 40 banks account for the other 50%. If you can get 2 of those 4 big banks embedded with long-term contracts, you can then provide something perhaps to the other banks that nobody else can provide to them because you now have this liquidity pool from the big guys or the trust or the brand or the whatever. And, and those, those types of things I think are durable once you, once you get those embeddedness combined with a marketplace. So again, you're combining different defensibilities in order to make each more sticky.

Appreciate going deeper on that one, James. So, hey, hey, Godfrey, do you want to jump on here? I saw you raise your hand.

Godfrey, okay, here. co-founder of Awele Technologies. We are connecting exporters and importers of used machineries to freight forwarders from US to overseas. Um, I'm following up on Jeff's question. James, um, with AI making it easier for products to be built, have you seen any changes in terms of how investors pricing technical co-founder for early-stage fundraising? Look, the investors are prizing Some very special technical founders if they're building AI tools, right? There's groups of 3 or 4 people who are raising money at a billion dollars because they're so good at technology, but they're actually making the AI. In terms of technical co-founders for building a marketplace, I don't know that they have been super valued for the last decade already. Because the building of a marketplace technically isn't that challenging. The challenges with marketplaces are getting liquidity. It's the order in which you get the nodes. It's your unfair advantage and your low CAC in terms of getting the harder side of the marketplace and then figuring out how to maintain that liquidity and keep loyalty. I think those things are not super technical. It's all just HTML and relational databases. The payments aren't that hard. It's literally just market construction, which has always been the challenge for investors when thinking about whether to give the money to a founder or not. It's not about their technical chops. I think for a decade at least, investors have just assumed that many teams are technically capable of building a marketplace.

Awesome. I'll try to get in on one last question that was added. And, you know, it was around the concept of the white-hot center, which, you know, we discussed on your past group chat. And, you know, they were asking, do you have any updates or maybe revisions to this, you know, for founders that are starting an AI Native marketplaces in 2026?

Unfortunately, I don't have any updates. I just, you know, this is a constant problem, which is, you know, you start somewhere and you think this is the best place to start because this is what your last job did, or this is what your last company did. And so, you know, this particular market, but the fact is somebody else has a different white-hot center, right? If you look at, you know, Facebook was the 50th social network and the white-hot center turned out to be Harvard University. And, you know, other social networks had 10 million, 20 million, 30 million users when Facebook launched, but they were in the right white-hot center. And figuring out what that is, you know, is challenging. And only someone who was at Harvard could have done it. And so for the founders who were the other 49 founders who had built social networks prior to it, they were SOL because they didn't have access to the most important white-hot center. So you need to be very curious about where the best place is to start on either side of your marketplace.

Great point. Way to wrap things up. So, you know, I know we're out of time here for the group chat, but, you know, really appreciate, you know, taking the time to join us for the follow-up. So we covered a lot, kind of packed it in. And as you could tell, you know, we're all kind of heads down trying to think about, you know, how AI is you know, not only I would say transforming marketplaces, but so many different opportunities as well. So this is a great conversation here and super helpful. And I had one, you know, my parting question that I wanted to, you know, reframe for this time, which is that if you could go right back to maybe like pre-2026, you know, what would you tell yourself about marketplaces?

I would say be very much more cautious about B2B marketplaces than I was. I've made a lot of investments in B2B marketplaces and it's been tough.

Well, I'll be sure to include, you know, links to your social and all, you know, NFX and a newsletter and everything so that way everyone can, you know, follow along if they're not already. But, you know, once again, really appreciate taking time to join us today for the group chat. And thanks everyone for joining in and the great questions.