AngelList CEO Avlok Kohli on the Transforming the Company — and Venture Itself

David: Hello, Acquired listeners. Welcome back to another awesome conversation here. I am so excited about this one today with Avlok Kohli, the CEO of AngelList. We were joking just before we hit record. Ben was like, you should do the intro on this one because AngelList literally changed your life. That's not an overstatement.

Just within the past few years, I and my friend, Nat Manning, have raised over $30 million together on the AngelList platform through a combination of funds and SPVs. We've invested in close to 100 companies. We've done so much. We are actually investors in AngelList itself. This is like a homecoming episode here. I'm so excited to do this.

Ben: Avlok, welcome to Acquired. Great to have you here.

Avlok: Thank you. Excited to be here.

Ben: The first thing that I wanted to ask you is around this concept of rebirth. AngelList is a 13-year-old company. If you would have asked me five years ago, I would have told you, yeah, I understand what they do pretty well. That's where startups that want to raise capital can go and find angels and do party rounds. I think it's for companies to go and find capital.

David: It's a list of angels.

Ben: And then, the business has completely transformed. My first question to you is, what happened over the last five years? And what is AngelList today?

Avlok: It's a great question. Early AngelList started off as an email list, literally, where a startup could come and meet a bunch of angels, raise investment, and go off and run the company. Uber was famously one of the first startups that did just this.

David: I hear there was some guy named Jason.

Avlok: Yeah. Uber actually raised successfully, including many other startups early on. For the first couple of years, AngelList was just that, and it continued to scale with just that. The theme to think about of early AngelList is it was really looking for product/market fit, and there was a team that was just focused on finding glimmers of product/market fit, like where is it? There was an issue around adverse selection.

Basically, if you announce to the world, hey, come here, you'll get money, you're going to get everyone in the world that's going to show up. You can imagine the quality of the startups starts to decline. At that point, there was an inflection point for the company around, how do you innovate out of this? What does that actual product look like?

That's actually how the original idea for syndicates was born. It was like, well, why don't we actually have GPs who can put skin in the game, then bring a syndicate of LPs, and they can actually serve as the quality filter, the signal from the noise? That was that original spark of that innovation that turned into syndicates.

Along the way, the early team—again, more of like a think tank—also focused on a few other problems that startups used to have. It's raising capital, it's hiring employees, and it's finding customers. Of course, syndicates were focused around raising capital. The job board, AngelList Talent, was around finding employees and hiring employees.

Product Hunt, which was acquired a few years later, was around finding customers. That takes us to the timeframe of 2018–2019 where you had these three different products that by this point, even though they solve for a single customer, so all products were a single customer, they basically branched out into several different product lines and really business lines at the end of the day.

On realizing that, the first thing that was done was to say, okay, let's experiment with spinning out one of the product lines as its own company because trying to do all three in one place is actually incredibly hard. It splits focus, splits the team, and all of that.

David: This is right around when you came in as CEO, right?

Avlok: Yeah. This was actually just a little bit before when I came in as CEO. AngelList Talent was actually the first to spin off as its own company. Even though it still shared the name AngelList, underneath the hood it was a separate corporate entity with a separate CEO with a separate team. That ended up working really well. That's around the time that I came in, which was in mid 2019.

The goal was to actually take venture, which is the financial platform, spin it out its own company, and then really think about venture as its own business and really rethinking it from the ground up of, if we're to build a venture scale business out of it, what does that actually mean, what decisions do we need to change, what do we need to make, and really focus on that.

The same thing happened to Product Hunt just a couple of years later. You can almost think about early AngelList as a think tank, and then it split off from these different business lines. When it comes to scaling a company, you really do need a team that's singularly focused on one thing, a board that is singularly focused on one thing. You can't have split focus on this.

The key juncture was actually that splitting off into different companies. Some of that, Naval and I talked about before I stepped it, around like, hey, if we're really going to make this thing huge like large business, what are the things that would need to be true, and how would we get there?

Some context on me. I've started three companies, one of them was bought by Square. For me, the only thing I really wanted to do was build something large. We really focused the entire conversation and discussion around, great, there's something that's obviously valuable in the venture business, and how do you take it? How do you build it into this large platform?

Ben: Now that Wellfound is the new name for AngelList Talent, the only thing that retains that original AngelList think tank name is AngelList Venture, correct?

Avlok: That's correct. AngelList Venture is now AngelList. Our purpose is to increase the rate of innovation in the world, so we've really embedded it.

David: We want to dive into a lot with you. I want to learn too on how this transformation happened. I know a little bit of it, but I just want to underscore and go back to something, like you said, a few minutes ago. The transformation and the brand perception—I know that's not exactly how you put it—of AngelList in 2017–2018 period before you took over in the transformation, if you had told me back then—I was a "professional venture capitalist" at large firms that went and made a decision to start a new fund—I would never start that on AngelList.

If you had told me that fast forward five years, and I would be running funds, managing capital on AngelList, and an investor in AngelList itself, I would have said, you are completely crazy. That is for there's the adverse selection problem, there's low quality stuff on there. I would never want to be associated with that. Today, it's completely the opposite, and I'm 100% in on the platform. This is an amazing transformation.

Avlok: I think the key reason for that early reception is because early AngelList did look like a toy. The SPV product, when you just think about that product itself, it's such a simple product, and it looks like a toy. If you are a professional VC, you're not really going to be using an SPV product. You're actually going to want to use a fund product. Not only that, but a very complex fund.

Ben: Avlok, maybe for folks in our audience who aren't investors or venture capitalists, why would SPVs look like a toy? What was it about them that felt overly simplistic?

Avlok: The way to think about a venture fund or a venture SPV is there is an agreement that governs all of it, which is called the limited partnership agreement. It's an LPA, that's the acronym. When you have an SPV, the LPA that governs an SPV is effectively a set of parameters. The number of parameters that you need to consider for an SPV LPA is, just to make a simple example, let's call it five parameters. Pretty simple.

Now, as you get into a fund, and you get into a larger fund, you can think of the number of parameters that you need to set up. It could be 40, it could be 50. It really depends. It gets very nuanced very, very quickly.

SPVs are simple. They look like a toy because there's just not that many parameters around it. A fund gets more and more complex because there are just many more parameters. The reason these products look like a toy and the reason AngelList only focused on SPVs in the very beginning, is because our entire approach to fund management has been distinctly different from anyone else doing this.

We truly are an end-of-one company, I have not found anyone else that's done what we've done. I think it's because we made an irrational decision early on. Maybe it's because we like doing hard stuff, or maybe we just didn't know what we didn't know. I've heard of many founders, CEOs who've built successful companies, if I knew then what I know now, I never would have done it. Honestly, I feel like that sometimes, where I'm like, yeah, I don't know if we would have done it back then, but we did it.

What we did differently from everyone else is we built software, and we took everything on about running a fund in an SPV. Okay, let's talk about software. What I mean by that is, we actually took the LPA. We wrote code to model the LPA in code, in a database, that then would output financial reporting, tax reporting, portfolio management, and all of it. We built code around, and we did all of it like many aspects of managing the lifecycle of an SPV and a fund.

When you're writing code to do this, you can't take on more complexity until you build a foundation. Early on, what happened was, we had SPVs, they were simple relative to a fund, we took the market on SPVs, and we actually became the market leader.

David: I think you really enabled it to. They didn't happen in the same way before AngelList.

Avlok: Exactly. You would need to go hire a lawyer. You would need to go hire an accounting firm. You would essentially pay a lot of money to do that. That actually, definitionally restricts the number of GPs that can use it. But when you build software for it, you bring the cost down.

By definition, you can bring the size of the SPV down, which then enables more GPs to get into business. What am I describing? I'm describing technology innovation at its core. This is what technology is. It provides more leverage to more people so more people can do it.

It's interesting. It's literally the classic innovation cycle. You start with something small, you really compress things that look like a toy, but then of course, you expand, then you expand, and then you expand. Of course, what happened was, we started SPVs that looked like a toy. But then once we had the foundational software built, then we expanded.

I think of this as stacking innovation. It's actually a concept I pulled from my time at Square, which is, as you're building the layers and you stack the layers, you have higher and higher defensibility. You can do more and more complex things because these innovation layers stack. They connect like Lego blocks, and then they effectively give you superpowers. They give your customers (more importantly) superpowers.

SPVs, we laid the foundation. Venture funds, we laid the foundation. A really cool thing is when Naval and I got to talk about rolling funds, which was a shower idea I had and connected this broader theme of, this is the way funds should be. I remember that initial conversation we had. He's like, yeah, I had brought this up in the past, but I don't think we have the infrastructure then.

Now that we looked at it, we're like, well, we have the layer that we built through SPVs, the layer we built through venture funds, and it turned out that those rolling funds, we had it all. We had the layers to now support rolling funds, which is an incredibly complex vehicle. We had the layers, and we actually went from idea to market in a couple of months because we had all the foundational layers built out, and the innovation just stacked along the way.

Going back to the earlier observation of 2017–2018, the feeling was around, hey, AngelList can't handle this, can't handle more complex funds. Honestly, that was valid. We couldn't, but that was the point. It was by design.

David: Did you even have a fund product at that point in time, or was it just the SPVs?

Avlok: It was primarily SPVs. We started working on funds. We were primarily focused on very small funds like half a million dollar funds, million dollar funds. If you actually map the fund size by year, you'll actually see a very clear increase in total fund size that AngelList is taking on. Again, this is the standard march of us just moving up market, moving up market, and moving up market.

Fast forward to today, we're supporting $100, $150, $200 million funds. Whereas two years ago, we would actually say no to these funds because we're like, look, we can't credibly support your fund for the next 10 years. AngelList takes an incredibly long view. We also take our commitment to our customers very seriously. I feel physical pain when our customers don't have a great experience. I really internalize it, and the team internalizes it.

David: I highlight too, you're building product and building software, but the product is a lot of money, people's livelihoods and company's livelihoods, which is also so funny given the brand and AngelList's origins and perception as this toy. But a large part of the Silicon Valley ecosystem now is managed by your product.

Ben: David, you've brought endowments on to AngelList by investing in Kindergarten.

David: Yeah, 100-plus-year-old university endowments, and now LPs of mine and others on AngelList.

Avlok: We're pretty private, overall, private in terms of what we share, but we have one of the highest quality endowments as an investor in AngelList as well. Again, this is all around, as we become the fabric of venture, we're essentially bringing in all of venture, all the large LPs, all the great GPs, all the best companies.

At this point, we're managing across 20,000 funds and syndicates, 13,000 portfolio companies. These aren't just records in a database where it's a read-only record, where once we never do anything with it. For a good portion of the funds, we're the signatory on the funds. We review the docs when the follow-on rounds happen. We handle the back-and-forth of the company when there's a distribution. And then we also handle the distribution to the LPs. We handle everything including banking. We actually have integrated banking that's built in.

We really do manage it all. When it comes to managing the portfolio, it is managing all aspects of that portfolio on behalf of the GPs. It really is the product that should have existed in the world but didn't because you just couldn't get the technology leverage back in the day. We really have absorbed all of the paper pushing, back office work that GPs don't want to be involved in because, again, a GP has a limited amount of time. They should really be focused on finding great founders, investing in great founders, and then helping great founders. That's the real differentiation with GPs, not back office work.

Ben: To grab a playbook theme that we talked about a lot here on Acquired, when you vertically integrate, you have to take on a lot of fixed costs, you have to do a lot of forward investments, but vertical integration almost always leads to a better customer experience. By you guys willing to take on all this vertical integration and all this platform years of building, it does enable this fast and fluid experience for GPs, founders, LPs.

David, correct me if I'm wrong, but when you and I invest in a company, and I have to go back and forth with the company's legal team like UCC and AngelList, I think they're the GP. You're not even technically the GP.

David: It's very funny. When Ben and I invested in companies together outside of the PSL remit, Ben, you're investing personally, [...], and I'm investing out of AngelList. Again, I never would have believed this. My life is easier than yours.

Ben: Is that right? Is AngelList actually the GP of your fund?

David: Yes, I don't even sign any documents.

Ben: That is serious vertical integration.

Avlok: Yeah, it's built on that. There are actually two different ways in which we can support a fund. One is, think of it as a full vertical integration where AngelList can be the GP if the GP wants that. That typically comes in when all the GP wants to do is make the investment, and then AngelList handles literally everything soup to nuts.

We also support the structure where the GP is the GP, AngelList is still managing the fund, and then there are aspects of it where we could be the signatory, and we would still review the docs. Think of it as a dial of control. You can have it in full vertical integration where you're like, you know what? AngelList, you handle all of it. I'm just going to focus on investing. And then you can dial up the control. The GP can say, you know what? I actually want more control. We now also support that.

Actually, I'm glad you brought that up because that was one of the other things that made it look like a toy earlier on. Early on, again, to reduce variability and to reduce the number of parameters so we can actually automate, build software, and build that foundational layer, we, by default, were the GP on everything, GP on SPVs, GP on funds.

David: I believe when I (in Kindergarten) started, that was the only option. There was no option not to be the GP, which was fine for us because this is our side investing activity. We're not trying to build a full-time firm here.

Avlok: Exactly. Again, we actually see also the split. There are some people that still are like, I just want AngelList to the GP, and people now just know this. LPs know this and they're comfortable with it. Again, this is the interesting about venture, and is one of the things that surprised me actually stepping in as CEO of AngelList is anything that's new or novel actually, originally is viewed as like, uh-uh. Not going to do it. I get it.

Early on, I was on some calls with LPs where AngelList was a GP and there were a lot of questions around it. It was like, why is AngelList a GP? Why isn't the GP the GP? This was part of the growing pains, if you will, of us building out the full product in the software stack.

But then, a couple of years ago we built an option. It's like, look, you can be the GP now, we're ready to do this. Of course that then allows for larger and larger funds to come on once where they do want more control, they do want to be the GP, and they want AngelList to manage everything else. That, of course, allowed us to continue to move up market, take more and more of the market.

David: Did that also unlock existing funds and firms already? For me, when I was starting Kindergarten with Nat, we were de novo, we were new. Of course, we're going to start on AngelList. But if you already have three funds under your belt, and you manage $500 million in capital, that's a much tougher decision to then move over to AngelList, right?

Avlok: Exactly. Another way to frame that is, we were literally creating the market. We're creating new GPs, enabling new GPs. As we continue to build out the software, the platform, we are now able to take on existing GPs. People have already started funds. People who do want to be the GP on the fund, people who need a lot more variability, a lot more configuration, a lot more customization. We've already built all of it because again, it goes back to stacking the innovation where once you lay the foundation, you lay the next foundation, lay the next foundation, you actually get compounding benefits.

There was actually a time, by the way, when AngelList had a waitlist because there were just so many funds trying to get in and we hadn't productized a portion of it. We just can't take these on. This is literally going to break us. It's going to break us.

I remember, at one point, I was dual heading the CEO role. I was also an account manager for 30 funds, launching funds. At night, I would go home, and I would get requests. Hey, I want to invest in this company. I'm like, great, let me work on it. I'm reading people's LPAs. That just to give you a sense of, there was a moment where we were absorbing so much demand, but we were still building out that software. But once it's there, boom, you get compounding benefits, compounding advantages.

We're now sitting on years of innovation that are stacked, that's compounded, that for us, we can just simply scale to increasing levels of complexity, scale to increasing levels of quantity, and we don't skip a beat. We're just moving forward smoothly managing all of venture.

Ben: It seems like there's got to be huge advantages to your ability to scale, having an engine that translates LPAs into code. Can you talk a little bit more about what that actually looks like?

Avlok: The way to think about the LPA, it's an agreement that binds the GPs and the LPs for the next 10 years. The agreement captures everything from who invests what, on what terms, what does the GP get, what does the LP get, what happens in the future around distributions. You can think of each of these clauses in an LPA has to be reflected in code for when there are specific events that happen.

One example of an event is the fund has to raise capital from LPs. Great. Now an LP is going to wire money in. What happens from the second the LP wires money in? You need to reconcile it, you need to match it to the LP. You need to take it, then you need to put it into the limited partnership vehicle, and then you need to make sure that ownership is captured in the correct way. That's the first part.

Next, what happens when you start deploying capital into companies? You need to capture the entire record of that. You just store that. Okay, what was invested? In what valuation? You need to make sure of the share price, the total number of shares is all fully captured. These events all need to be managed.

I share this with the team internally where I say, look, we can't be 98% accurate on things. This is a financial instrument. If your bank was 98% accurate on your cash that's sitting there...

David: Too soon, Avlok. Too soon.

Avlok: Yeah, that's right. Exactly. Technically, I think the FDIC posted 81% or 88%.

Ben: Yes. If your bank is 98% accurate, or if I make an investment in the company, I wire $3 million in, and they're like, great, we got 2.95-ish, give or take 10%, I'd be like, okay, this is 0% accurate. This is not 98%, 0% accurate.

Avlok: Exactly. When you think about all of the events that have to be captured precisely and correctly across the entire lifecycle, 10 years. It's not, oh, you do it for one year and you're good. Nope, 10 years. These are obligations.

Everything is captured across 10 years. It has to be reported around financial reporting, around tax reporting, the distributions. Having everything captured correctly and accurately for 10 years, ends up being really, really important and can actually only be done with a platform play. It's incredibly hard to do it in the way that the rest of the industry does today.

This has been the other interesting part about running AngelList. As we're building, it's just becoming clearer and clearer that this is the only way that this industry will scale. This is the only way that people can actually build and scale venture funds because the other way just has a lot of human error, a lot of data portability issues.

Most other providers only support one small slice of it. You'll have someone that only handles subscription documents. You'll have one vendor who only does financial reporting and accounting, another one only does tax reporting, and then another one that only does investor relations.

Our view has always been, we should be able to do all of it. That's really the ideal product. That's what GPs really, really want. In fact, by not doing it, you're actually keeping a lot of GPs out. People who could be great investors are not getting in because you know what? This is too much back office overhead.

David: Totally. Nothing gets these firms. They're great firms. You walk into a Sequoia, you walk into an Andreeseen. You walk into any venture firm that, in recent years, has been managing large amounts of capital across multiple sectors and stages. There are 40, 50, maybe 100 people working there in the back office managing all of this. The GPs didn't get into the business to do that.

Avlok: Exactly. Actually, we're managing it very privately, I can't share the name specifically, but I can share the other firm. We're managing one of the top 10 VC firms. We're managing the entire scout infrastructure, all of it, end-to-end, the entire thing. It is actually a result of us building this platform that can handle more and more complex funds and volume as well. We're handling it end-to-end, every single aspect, including compliance, portfolio management, scout management, literally everything. By the way, it's all run through software.

David: Not to preview your customer conversations over the next few years, but that's like a gateway drug too, right? Like, oh, manage my scout program. Five years from now, I'll be managing the whole firm.

Avlok: Exactly. This is just a classic move up market continuing to take more and more of it.

Ben: Avlok, I want to keep digging at this LPA to code thing. You could imagine, it's pretty easy to translate some parameters of LPA to code. For example, oh, I don't want 2%. I want 2.5%. Or, I want 2.5% at the beginning, 1.5% at the end, and I want each year to incrementally change between those two things.

Let's say I get really clever in my LPA, and I say something like, well, I want, of course, an escalating amount of carry depending on how much capital has been returned so far, and also I want a non standard window. Let's say I distribute shares of a company after it's gone public, I don't want a five-day look back at the average price of the S&P 500 in terms of what I get my credit for. I want a 20-day look back. My LPAs agreed to it, and it signed a signed document. How does wonky stuff get translated to code?

Avlok: In a few ways, first is having an infrastructure that is flexible enough, where you can actually stack and add these different types of terms so that it isn't fixated on, okay, you can only do this particular term on look back. You can only do this one look back. You can actually have the infrastructure itself be flexible.

Second, there is an aspect of reading the legal documents and then translate it into a mathematical representation of what the clause means. Legal agreements are basically code in English. That's what it is if the agreement's written in English. The job to be done is, how do you translate that English, that clause into some formula that Ben can get represented in software? As long as the infrastructure supports that, now the job to be done is, how do you translate that LPA and the clauses of the LPA?

This is where, as we're making big advancements in AI and large language models. What is a large language model? It is using English, a natural language, as a programming language to query other natural languages like other unstructured data, so you can effectively ask unstructured data to structure itself, and that's where you get back.

We have a lot of innovation inside the company that's focused just on that. We have a huge repository of LPAs, obviously, because we're supporting so many of them. We've already, through humans, taken the LPAs and represented them in code today. We have this repository that we can then feed in, such that going forward, we can actually have the LLMs prompting as a way to pull the necessary information out in representing code.

By the way, that's just one small aspect of what it means to build software for a venture fund. There are many other aspects as well around managing it for the full lifecycle, as I shared earlier.

David: Oh, my God. I'm just thinking of tax, K-1s.

Avlok: Everything. Oh, my God.

David: What a nightmare.

Avlok: I don't think if this has been shared publicly before. There are a hundred workflows that we execute on, on behalf of all the funds in SPVs. These are a hundred workflows that are operational workflows. That could be anything from updating portfolio valuations, preparing K-1s, responding to investor questions about K-1s, a hundred of them. That's a hundred that all needs to be done precisely and perfectly for 10 years. Of course, it's a rolling 10 years every time you're taking your fund, a 10-year obligation.

What's actually been happening internally is, bit by bit, we're actually turning the non-engineers in the company into engineers using LLMs. We have an in-house tool that is effectively the operational nerve center that we've actually hooked up to GPT. It's like a flowcharting tool. You can pipe different parts of the workflow, and then you hook it up to GPT for different parts.

You can say, for this set of emails that come in, automatically categorize all the emails. Okay, great. This one's a follow-on round for a Series A, this one's a new investment, this one is a tax question from an investor. Okay, great. Based on that, go down a certain path.

Let's say it's a Series A follow-on round that we need to review the docs and make sure all the investor rights are managed, make sure the cap table calculations are correct. Great. Read all the documents automatically, by the way. Categorize all the documents. Okay, we've got the cap table, we've got the share purchase agreement. We know that for every Series A follow-on document, we need X number of documents and select the cap tables missing. Automatically, we see the cap tables missing. Draft an email, autogenerate it, mind you.

David: It's so funny. You're saying this, and I'm like, oh, my God, I've been living this for the past several months. I know everything you're saying. I'm like, yup, I've experienced all of these.

Ben: Like you're on the other side of these emails?

David: Yeah. We just had one of our portfolio companies go public. We had some questions back and forth with the AngelList team. I've started noticing on that and another one, you guys have been proactively suggesting, messaging in bullet points for us to tell our LPs. Now I'm like, I see where this is coming from. This is amazing.

Avlok: Yup. Again, the reason we're able to do this is because we've already been doing this for a long time. We have this massive set of training data. It's the best way you can use this data to fine tune.

For example, going back to the Series A follow-on round, let's say the cap table is missing. Great. Hey, can we get the cap table automatically? Great. Now we get the cap table. Ingest the cap table through code, through LLM. Ingest it, structure it. By the way, we've already built tools for non-engineers in the company to build all of these automations for their workflows, and we're seeing a lot of movement. It's actually fascinating.

At AngelList Confidential, which is our annual conference, I actually did a fireside chat with Sam, and that's what opened my eyes. He said something to me that just stuck with me, which was that the cost of intelligence is going to zero. That just stuck so fiercely in my head that I basically took that and I was like, huh. Things that we assumed a human needed to do, they don't actually need to do anymore. If we can build tools inside the company for non-engineers to use, such that they can now take intelligence and automate it because the cost of intelligence is going to zero, we'll get insane leverage, insane automation coming out.

David: This is crazy. This is going to have such a transformative effect on you as a business. I remember when we were investing in you guys. We got questions like, isn't that a super heavy services-oriented business? As AngelList scales with funds, with companies, with everything, they need people to service all this stuff. This is a major unlock for you guys, right?

Avlok: Exactly. All you need to do is follow that cost of intelligence curve. The way to think about it is, as long as the tools are getting built out, such that humans can take their intelligence and automate that intelligence, because again, the cost of intelligence is going down to zero, you're going to see an insane amount of unlock, insane amount of leverage, and insane amount of automation. Of course, the companies that are best suited to take advantage of it are the ones who have a depth of experience, depth of training data, in order to be able to do it well. Those will win out quite a bit.

David: The vertically integrated with good established software foundations like you guys.

Avlok: Exactly. You can then just imagine a hundred workflows with multiple agents per workflow. It's just a concept of how you chain all these different tasks and workflow and you automate different parts of it. Basically, think about it as all these bots that are running AngelList operations.

If you visualize it, just visualize we have a few bots today, we're going to have many bots in the future. Each of these bots are highly intelligent bots that are very, very good at a specific thing. They're essentially helping run the entire world of venture capital. That's the way we think about it. That's what gets us incredibly excited.

David: It's so funny. Without realizing it, I'm like, oh, yeah. I've been on the other end of this for the last few months. I really have noticed a change in the nature of interactions with AngelList, the types of responses, and especially the proactiveness coming out of the platform. Now I'm just like, oh, of course, it's LLMs.

Ben: There's an interesting high level trend that you're hitting on here because five years ago, fund-backed offices were exclusively a low margin services business. Through all the software you've built and then this new edition of LLMs on top, you could really imagine it becoming a true 80% gross margin SaaS business. It's this unlock of, oh, what other industries are perceived to be only possible to have services business margins, but actually could have software margins?

Avlok: Bingo. The assumption that all services businesses have low margins comes from a world where the cost of human intelligence was high, because there were things that computers just couldn't do, and there were things that required engineers to tell computers what to do. Let's cover both.

There are some things that computers could not do. Computers could not solve these read an LPA. They can solve these unstructured problems or these fuzzy problems. They need a very precise problem to solve. Great, massive unlock. You can now solve fuzzy problems, things that you actually thought you needed a human to do. It's like a lot of pattern matching base, a very complex pattern matching.

The second big unlock was, it used to be the case that you needed engineers to tell a computer what to do. Guess what's happened, English has turned into a programming language. You don't need an engineer to tell a computer what to do.

By the way, there's a third big thing that is a huge unlock. There is no need for hardware adoption. Everyone already has a supercomputer in their hands. You have 7 billion people in the world that are all connected on the Internet.

Take all of those three combined, 7 billion people connected with a supercomputer in their hands, they can tell a computer what to do using English, and the computer can do what they want it to do and solve the fuzzy problems, the cost of intelligence is not high anymore. Things that were typically limited by service margins, they won't anymore.

David: Also, it's not high and it's also not linear, right?

Avlok: Yup. I think there may be a different problem to contend with now, which really may be differentiation. How do you now differentiate? Maybe that's what eventually drives costs down or the margins down, but that's where if you have specific moats, you're vertically integrated, or you're very specialized, you can actually have high margins. But I think the assumption that you need humans to do things and margins are low is an assumption of the old world, not the new one.

By the way, the new one is coming fast. This is coming very, very, very fast. We add our all hands today. Every Tuesday, we do all hands. One of the teams that's been working with a lot of the internal tools that we have hooked up to GPT was sharing an automation they did. Everyone just went, wow, that's insane. It's coming so fast to all other companies.

We're not the only unique company that's doing this in terms of looking at all the different parts that we can automate inside a company. You can imagine all the different use cases that are getting deployed very, very quickly. We're in the midst of the biggest technology wave, the biggest, 100%.

David: If you're willing to share, how many people work at AngelList right now?

Avlok: We're at 170-ish total. That includes everything, including our back office, AngelList, any affiliated entities we have. It just includes everything, it's like 170.

David: Already today, the amount of capital that you manage across any dimension—dollars, number of entities, the number of investments, whatever you want to take—there's no way that only 170 people can manage that in the old way. Zero chance. Andreessen has more than 170 people working there.

Avlok: We shipped 350,000 K-1s last year. It's probably going to surpass that this year. You can't do that through humans sitting there just drafting these K-1s and sending them out.

David: Totally. Kindergarten alone, I think, is responsible for 100 of those.

Avlok: Yeah, there is a significant amount of volume. AngelList really is at its heart. We're just a bunch of tinkerers and builders. Truly, it is founders building for founders.

I basically got the build tools that I would use. I actually use every single product of AngelList, except for the cap table, I need to start a company. I was joking around like, I just have to start a company so I can use the last product I can't use right now, which is the cap table.

David: The AngelList cap table isn't on AngelList?

Avlok: No, that one, but I'm talking about myself as going through as a net new company. I'm like, I just really want to use this cap table product. The one thing that is incredibly interesting about this moment and also about the correction that we went through around a lot of the layoffs in the tech industry, was really all of us finding our footing and ground of a technology company is meant to scale and increase the leverage per person, not scale headcount based on how the company is scaling.

The way we've run AngelList has always been, how do we just bring together the brightest individuals who can really use leverage? What really are the forms of leverage? You have code capital humans. Humans are the worst form of leverage because leverage is a function of the humans you hire, so we really leaned pretty strongly on code. How can we actually get higher leverage per person? That's how we've actually designed the company.

We have different ways in which we have that force, function, and the constraint. One of the most recent ones is, this has been the case for a while. I've basically said, hey, we're going to keep our headcount constrained. We're still hiring new people. That's natural as we're backfilling roles.

My message has been pretty clear. We're going to keep headcount constrained. We're going to force the decisions needed, such that we're building more and more and more leverage to the business. This is something I actually picked up from early Square.

Square was actually ahead of a few things on a lot of different trends. One of them was when I first was acquired by Square. The first message we received from Jack was, all right, I'm not adding any more people in the company. It's a good forcing function for the company to really focus on the points of leverage and building out those points of leverage.

There are all these things that I actually bring into the company and test around and play around. One of the other ones I love doing is I'll just cancel all my meetings every quarter and see what happens. If it's important, I'll get back on. Sometimes I'll just cancel one-on-ones entirely.

David: I love it. It's like garbage collection for your schedule.

Avlok: Exactly. This idea that there's a calendar invite that was set six months ago is even relevant in terms of the prioritization of today, makes no sense to me. It's really good to try out different things that help shock the system because that's really where, I think, teams and individuals can actually explore new things. Again, going back to increasing leverage for the company, there are things that we do that continuously test that and push on it. We take that very, very seriously. We really do believe in individual exceptionalism; that's another thing.

I really expect that any person that joins the company can create a new founding moment for the company. They can truly change the trajectory of the company because I've been in and around founders for the better part of the last decade. I've actually seen how a single individual can just change the world. I do believe everyone is capable of it. That's another thing that we really believe in at AngelList.

David: It's also like your whole business. That's the customers, but that's the ecosystem you serve. You have the data right there.

Avlok: Exact. It's literally in our ethos. We truly do believe a single individual can change the world. We're serving them. We're actually serving the GPs who go back those individuals are going to change the world. It is a little ingrained in our ethos.

Ben: As listeners think about the business of AngelList today, which is simplified, Wellfound has been spun out so there's no more talent. You mentioned Product Hunt, that's been spun out.

Avlok: That's correct.

Ben: The business of AngelList today, how does it break down? You don't have to give specific numbers, but how does AngelList make money?

Avlok: There are three customers of AngelList. The first one are the GPs. These are the folks who start SPVs, start funds. The second are the LPs. These are folks who invest behind the GPs. The last one are founders and startups.

For the GPs, two products, SPVs' funds, pretty simple. For LPs, they're typically investing behind the GPs, or sometimes they'll actually find other deals to invest in on the platform that is a part of the business, but the core part of the business really comes from the GPs in terms of starting SPVs and funds.

The last one for founders and startups, really, it's fundraising tools and cap table management. We actually launched cap tables last year. It's actually growing very, very quickly. We're incredibly excited about the rate of adoption on that product. Each one of these products has their own revenue model. For SPVs and funds, we charge an administrative fee to manage it over 10 years.

David: Can I just double click real quick on that? We didn't have a choice. Our first Kindergarten fund was $3 million. There was no other way to do it except on AngelList. But then as we got bigger and managed more, having the experience of having started a firm before I made all the platform decisions, the cost of going with you guys to do that for the administrative fee, is night and day. It's a 10X difference versus doing it the traditional way.

Avlok: Exactly. Again, it goes back to because everything is vertically integrated in a single place, we can actually absorb all the different service providers in one and then provide a single price. Otherwise, you need to go and stitch together, and everyone's going to quote you something different. You want tax provider, fund, admin, et cetera. That's on the GP side. That's how we make money.

On the LP side, if there's a primary investment or secondary investment, if there's a primary investment or secondary investment that AngelList has helped facilitate and bring to the table, this is what we call AngelList LPs, then there's a broker's fee on that one.

Ben: Where you're matching supply to demand between LPs and GPs.

Avlok: Exactly. By the way, you can actually see where that goes. The product strategy goes there around, look, today, a lot of it is primary investment, but we're doing a lot of very interesting experimentation around how do we help funds provide their LPs' liquidity within the fund itself? This already happens offline today, and we think we have a unique approach there. We're going to be launching a product around that pretty soon. That's where GPs control and manage. That's on the LP side.

David: Would that look something, I'm just imagining, to Kindergarten, we have stakes in close to a hundred companies, many of them now are very well-known, lots of people, would like to invest in those? Some of our LPs may want liquidity. AngelList would be a platform to manage that.

Avlok: Exactly. We can be the platform that manages that end-to-end. We have a unique view on it as well. We think that we can help facilitate that in a single company in a portfolio as well. Again, once you get into the nuts and bolts of the LPA and the way you can—

David: Which in the old world, the only way that happened was, let's say I was Kindergarten, that needed to happen. I sold a piece of the whole commitment into Kindergarten, the whole blended fund, but you can atomize it to a company basis within a fund.

Avlok: Yup, we have a unique approach on how to do that and how to make that work. On the founder and startup side, the revenue model is a SaaS fee. Pretty straightforward—you're managing the cap tables, we charge a SaaS fee. It depends on your size of company and the number of investors you're managing on the cap table.

Then, there are other fees that get baked in because once you're a financial platform and you're moving money in and out of a financial platform, there are other different ancillary fees and everything that you can actually earn by reducing more and more friction. That's the other piece that's not as well understood about AngelList.

As a platform, it serves three different customers that have a portfolio of products. There are actually many revenue models that you can earn, especially as you get larger and larger. A lot of what we're doing is just continuing to build out this platform.

We're actually going to be launching a series of products over the next 2–3 months that will further emphasize what product strategy we have and how we're going to blanket the rest of venture. We're actually taking a lot of what we've built over the years. We're going to start unbundling pieces of it. You'll see how all of it is going to connect. It's a continuation of our strategy to move up market.

I'm very, very excited about this. Our team is just pretty jazzed building this out. It's a part of the evolution of a platform. That's another revenue model that starts coming in as well.

Ben: On the breakdown question, I have to imagine, SPVs were a large part of the business before. It seems like the funds business has to be meaningfully larger today than SPVs ever was.

Avlok: It is. Even before the market crash last year, May of 2022, the SPV business was not the largest part of the business. That's actually a fact that most people may not know, but SPVs haven't been the largest part of the business for a while now. The funds part of the business is definitely the largest part by many multiples.

Ben: Interesting.

Avlok: Very interesting. Some people may think of SPVs as the key thing. Don't get me wrong, we're going to continue to support SPVs. We're still the market leader in SPVs. It's a very important component of our product portfolio. But AngelList is firmly in the funds business as well and actually has a very strong stronghold in the funds business.

Each time we go into a new target in terms of fund size as we keep moving up market, if you map our market share each time we enter a new fund size, we essentially take the market each time. We just fundamentally just absorb it, take it, then we move on, and then move to the next one, and the next one. Just by virtue of us building that out, the fund side of the business is just much larger.

David: There's something that's in my mind on the SPV, and fund dynamic has been the case. I don't know if it's just my experience or if this also reflects the broader market, which you have the purview of. To me, SPVs have really changed over the last five years from being a way for individuals and groups to invest in companies that otherwise they wouldn't be able to, to more of an elastic scaling product for funds, or at least that's how I think about it.

We have core funds that we invest out of, that are blind pools. Then when we have opportunities to write larger checks, we use SPVs to flex up on certain investments. Is that the primary use case of them now or is that just my use case of them?

Avlok: I would say that's one of the use cases, where a fund uses an SPV to amplify the check they could be writing into a company. The other use case are GPs who are just running SPVs, and not writing checks from their funds.

The SPV business segments into two types of GPs, one who are doing exactly what you said, amplifying checks from the fund, and then the other GPs are only doing SPVs as the primary thing. The use case for both actually looks very, very different. Anyone who's amplifying from the fund, typically is really looking to bring their LPs in on an opportunity where they can invest behind the company if you can get a larger allocation.

The ones who are only investing in SPVs are really optimizing around AngelList LPs, how can you bring more dollars into a company. It really bifurcates into the two, and they do look very, very different.

David: Those two different types of SPVs and what we were just talking about there, to me—that's a good way to start to wrap things up here—it's really changed my thinking on what a venture firm and fund can and should be in the pre-AngelList world, pre-atomized, pre-softwareatized world of this. You needed a certain fund size to just be able to do things.

Your strategy was literally limited and dictated by the size of your fund. That's now no longer the case because whatever your core fund size is, or at least this is how I think about it, you still can do much larger in different things by amplifying via SPVs. The complexity around that is enormous, but I think it's really interesting. It's, in many ways, a better product as I think about it for all the participants in the ecosystem as a GP.

I can get deal-by-deal economics on my high conviction. That's where I have access as an LP. I don't have to commit tons of capital to this blind pool, or I don't know what's going to happen with it. As a company and a founder, I still can go. These days, I am going to the large established funds to be my leads and all my rounds, but I can get a lot of capital from these other folks that I otherwise couldn't get.

Avlok: What we're noticing as well, if we rewind back to a few years ago, is that founders weren't as familiar with SPVs. The world of venture has always been this very mystical to founders. If we fast forward to today, with AngelList RUVs (roll-up vehicles) which are fundraising tools, we've normalized this idea of a vehicle where you can bring many checks in, and it's a single line item on the cap table.

What we've noticed is because of that, more founders have been educated out of the gate on what SPVs are and also more and more open SPVs if you just crack it over time. I think there's also been a shift in the Overton window around SPVs, understanding roll-up vehicles. All of this coming together actually does firmly go into the world and lead into a world, where founders are going to be more and more willing to take smaller and smaller checks wrapped up into an SPV or wrapped up into a roll-up vehicle and really get the supporters to help them build the company. I think that's incredibly important.

We're seeing this across the board. We're actually regularly seeing large established venture funds, educating founders to go to an RUV because they're saying, use this as a way to bring on helpful operators and investors into a simple cap table line item. I think we're firmly in this world of, how do you bring on many more people to help support the company so you can continue to build your company?

Ben: Nothing better exemplifies your long-term vision than that. Avlok, where can people find you on the Internet? If they're interested in any of the stuff you mentioned, where can they go to learn more about AngelList?

Avlok: To find me, you can follow me on Twitter, twitter.com/avlok. I'm always there. To learn more about AngelList, you can go to www.angellist.com.

David: Great. If folks are interested in becoming GPs on AngelList, what does that process look like?

Avlok: Really fast. You just submit the form online, we get back to them within hours, we go through a quick call just to get a better understanding of the fund, and we then move forward to the next steps.

Ben: Technology, it's going to change the world.

David: So great. Thanks, Avlok.

Ben: Thanks, Avlok.

Avlok: Cool. Thank you.

Ben: Listeners, we'll see you next time.