Episode #23: Unique Revenue Channels For Vertical AI, Ghosting in Venture, Is Data Driven VC Over?

We're in a moment where just the budgets are big and I think no CEO wants to go to the board and say we're not heavily invested in figuring out how AI is going to impact our business. At some point they're going to have the answer. They're going to make. AI right now impacts this part of the business. So we're going to spend money here. We tried this thing, it didn't work. We're not going to continue to spend on that and it's going to settle out. But right now it does feel like a gold rush.

It's the week of March 10th. Welcome to the Learning Corner, a Precursor Ventures podcast.

Welcome to the Learning Corner by Precursor. I'm Mia Farnam and I'm joined by our GP and Managing Partner Charles Hudson, where we will be talking through the articles and Goodreads that caught our team's attention over the past week. Our first topic is from Tech Rundown by Caroline Thompson. Double clicking into the recent 300 million Series D led by Sequoia into Harvey AI, the legal tech startup now at a valuation north of $3 billion. What makes this deal unique isn't just the size of the check, it's the strategy. As a three year old startup, Harvey has nabbed partnerships with massive professional services companies to create white label tools to support their client work, all built on Harvey's AI. PwC has secured exclusive rights to Harvey's models and law firms are offering their white labeled product as a premium service billed to clients under Harvey AI's tech. Because these professional services companies are opting to resell Harvey's TEC through their own revenue channels, Harvey both sidesteps longer enterprise sales cycles and secures reoccurring revenue without having to build out a massive direct sales force. So I was curious to know what you thought about that one.

I'm always hesitant to comment too much about a company where I'm not on the inside, but a couple of things I remember five or six years ago, everyone was like, don't sell the lawyers. They're like the worst customers ever. And the more I read that article I was like, this sounds kind of like Palantir and more like a platform and a set of capabilities. Because I've always wondered like, what exactly is Harvey? And I've asked my friends Law and the few friends I have who've used it, when they describe their product, it's like they're not talking about the same thing, they're all using it differently. I kept reading, I was like, oh, this sounds like A really powerful set of capabilities that also, to the benefit of the consultants, need to be customized and integrated with your workflow. And it wasn't so long ago that people were like, services equals bad, software, revenue equals good. And in some ways you kind of get the best of both worlds. If you're Harvey, you've got a technology platform, you've got a service delivery partner who's going to help integrate it. But my big takeaway is that to me would suggest that Harvey will end up being pretty sticky in these law firms, because every other thing I can think of, whether it's hr, information systems, CRM, anything where you bring in a big integrator, usually ends up being highly customized, expensive to maintain and hard to rip out. And so I'm very curious to see more about like, what these Harvey implementations actually look like. But it sounds like they're moving in a direction that to me feels more like a Palantir or a success factors or one of these like system of record type things that requires a lot of customization. I was like, it's like a thin wrapper. I'm like, it doesn't sound like a thin wrapper if you've got those folks.

Involved, especially when you think about the consulting side of it. So when PwC kind of steps in, it's no longer a wrapper, it's a full blown product that they're offering, but under the guise of their own marketing. And so to your point about stickiness, that's something that you would assume will never leave their offerings as long as it continues to produce some sort of ROI for them and their end clients. That was actually my next question. We always talk about how AI is interacting with different fields and in this case it's professional services. So we're seeing here that big players in this space are admitting they can't outpace the tech. So what they're doing now is looking for ways to leverage it. Does this speak more volumes about the advancements and benefits you gain through vertical AI? Or how perhaps there has always been this cap on value for professional services at large?

One of our LPs is a management consultant who does a lot of these integrations. And last year I was giving him a hard time with AI. He's like, this is the greatest thing to happen to my business in forever. He's like, we have this new technology that my customer doesn't understand but desperately needs to use. I go, when was the last time this happened? He's like, the Internet. We had all these people who were like, we need a website or we need an E commerce capability and the tools weren't sophisticated enough that you could do it at the scale that a Macy's or a McDonald's or whoever would need it. So you bring in the consultants to help you do the Gen1 integration and build. And that also tells me a lot to your point about probably the state of the technology being so new you probably do need someone to come in. While you could probably get three engineers over a weekend to build you an mvp, you couldn't build something that an entire so I think it's really interesting that the professional services folks have found what sounds like huge new channel for both consulting services and you know, well pilot projects and real sounds like sticky deployments using this AI technology and we'll see if it goes the way of the Internet in five or seven years. Some of these things that you had to build custom with a partner, you'll be able to buy a standalone product. So I'm very curious to see how quickly that evolves.

Yeah, and it's not entirely new. The business model of teaming up with another company to extend out services I think of like workday offerings and how that exists in every single consulting firm, SAP, Oracle, all of those. What's different or feels different is leveraging Harvey AI's tech and white labeling it as your own proprietary, trying to get exclusive rights to the other Big four can't have their take at it. That's what feels a little bit different. And I wonder if it's because they know that it's difficult to build that sort of tooling in house. So why don't I just partner with the best of the best and try and edge out everyone else? Our second topic is a TechCrunch article by Marianne Azevedo digging into a long standing frustration for founders. Why do some VCs completely disappear after an initial meeting or eventually reject a deal even when the conversation seemed promising? While investors don't own founders capital, the lack of communication oftentimes leaves many feeling confused about what went wrong. We had a conversation around this from a piece Chris Newman had written a few months ago, and the article echoes a lot of the same possible reasons ghosting happens, so lack of conviction, internal dynamics, timing, etc. Ultimately we all agree that these reasons don't make ghosting right, but at least we can shed more light onto what might be the reasons you never hear back. One thing that we didn't chat through in an earlier episode were common red flags that investors might see that perhaps even before Taking a meeting could really dissuade them from taking a call or responding to an email at all. One that was mentioned here. It hits so close to home but it AI based outreach and note crafting. It's becoming so easy to blast off your deck and a blurb to as many people at once and hope by just pure volume that you'll hear a response. I will go on the record and say it's very obvious still when you are doing that and for me not the best or most effective way to get an intro to an investor. So I wanted to hear your thoughts on this as well.

Yeah, Mercedes at Lightspeed also had a very brave post that I would encourage people to read the comments. I remember in the arms race of VC attention, the first volley was really things like sales loss and outreach. Just sequences. People were running sequences. I'd get an email and then I'd open open it and then I'd immediately get a hey like notice you checked out my deck or hey like just wanted to follow up. So I got used to these programmatic sequences. But most or at least a good chunk of those emails were written by a human. At least most of it was written by a human. And then I feel like we now have this data pipeline where you can go on LinkedIn. You can get these airtables of like thousands of VCs with their contact info. You can hook that up to ChatGPT or Cloud or your LLM of choice. You can get some delivery sequencing and now you can have personalized turbocharged. And I've looked at some of these GTM stacks. There's a bunch of tools that will do enrichment so you can give it all the emails. It'll enrich some, it'll pull out some fun facts and you could push go. And what do I owe that person? Yeah, I don't think I actually owe that person anything. We haven't interacted. I haven't committed to meeting with you. I don't think I owe that person anything other than the time I spend opening the email, making a decision whether I want to read it. I think the bar does go up when you meet with somebody either in zoom or in person. I do feel like I owe those people and I keep tabs on it. I'm in like high 90s closing the loop. I will even sometimes I'm embarrassed to say I this weekend I wrote a bunch of emails to people that I just had dropped the ball on closing loop with them like two months ago. And I was like, I'm sure You close your round. I just want to apologize. It was rude of me to not get back to you. I should have done better. I hold myself to a higher standard, but the volume of inbound stuff, it's actually over personal. The ones that stick to me, I'm like, you don't know that much about me. And there's the only way you would possibly know that much about one line about, oh, you're a Chelsea football fan or, oh, you, that's cool. Three paragraphs about that stuff. I'm like, this is. I mean, I don't know. What's your experience?

No, yeah, I mean, it's the same. It's the same. I mean, we were just talking about it. I think the use of AI tools to make certain manual tasks a lot more efficient for yourself is a wonderful use. When I think it's crossed the line, I think is when it's just adding fluff to, I don't know, mask the ultimate reason why you're sending an email. And I don't look down at all upon people who inbound me with no connection whatsoever. I read every single email that comes through my inbox. But now I'm finding I'm actually paying attention a lot more to the emails where it's very quick, very straight to the point. You have the information there, a very clear way to reach you. If I want to learn a little bit more, it becomes very clear that it's not actually personalized when it's five paragraphs deep with all these fun facts about myself. So I would just say it's almost like a reallocation of your effort when you're thinking about those emails. Maybe that's just not the best use of your time.

I also think, like, in all sales situations, people get ghosted. LPs ghost me, investors, ghost me, investor.

Trying to be that happens.

And it's hard to tell. People don't take it personal when it's your company and your vision and you're putting yourself out there. So I don't tell people, don't take it personal. Of course you're going to take it personal. And like, just as a different way of saying no, I thought like, hey, if I haven't gotten back to you and you make an extra effort to reach out to me, that gets my attention. If we met and I said, I'm going to get back to you in a week, 10 days, and it's been two weeks, we haven't heard from you, are 100% entitled to nudge me. And as the post said. Sometimes I'm like, hey, it's the end of the quarter for audit season. The only thing I can think about is like, audit and your company could be super interesting. Or I have a portfolio company that's going through something dramatic. Everything is kind of behind that. So it's not always like you. But the thing I tell our portfolio companies is when you are one of the most important things happening at a firm, you will know. Responses are quick. Resources are like, you will know. And if you don't feel that energy, then you should read that. As I am not high enough on the priority list of this firm, it could be something that I haven't done in getting them excited. It could also be something internal to that firm that's going I had someone I was trying to chase down who was normally responsive and who sort of ghosted me on three VC intros and he's leaving his farm and he's like, I didn't really want to talk about the decision until it was done, but I also didn't want to waste those founders time taking intros. And I was like, oh, I was beginning to worry it was three in a row. So I don't know. I really do feel badly when people have made the time to walk me through what they want to build. I mean, one other thing, Mia, is I feel like back in the day a pass was just a past. You're like, we're not moving forward. There's no excuse for not today. But now I feel like the standard is your decision plus some rationale. And sometimes it's hard to put into words like, well, why am I passing? It wasn't one thing. It was like the combination of these three or four things not aligning. And sometimes I get I'm staring at the draft and just saying what goes in the box.

Tying this all together in the thread of AI that we've continuously talked about. Another red flag worth noting from the article is exaggeration and quote unquote dishonesty around the traction for the business that could signal someone to pass, even if you've had a wonderful set of meetings. I'm finding in AI, some of this quote unquote dishonesty isn't really 100% the fault of the founder at all. I think that we're just going to have to get used to seeing a lot of more false positives, specifically with an AI startup so early on in their journeys. You shared a great post by Samir Kaji, who shared the term annual curiosity revenue or capital burned by companies so potential customers as a mechanism to try out AI tools only to realize there is really not a clear ROI and then churns as a result. So we both laughed about this one. So I was curious to ask you about your opinions here as well.

I had someone who told me they're on a $10 million run rate. I was like, that's amazing. They said, yes, if you multiply our best week on the revenue run rate by 52. I was like, that's not. I was thinking you were doing more like, I don't know, 800,000, 900,000amonth. Not like you had one amazing week. And I think AI has distorted people's view because you hear these stories about companies going to 10 million in ARR in six weeks, 100 million in ARR in two quarters. And I don't know. I've always been a believer that if one company has standout performance, it's probably a special company. If a dozen companies have the same kind of vertical growth like this in a compressed period of time, there's probably something macro happening. And my belief is that annual curiosity revenue is a pretty. I wish I'd been at that dinner and no one had said it, because I do think there's a lot of companies who are just like, we're going to try as many AI tools as we can in search of the thing that works. And is it very possible the thing you bought last quarter is no longer the thing that works because some new capabilities come out? And so you try a new thing because nobody wants to get left behind? At some point it's going to settle. People going to be like, all right, I'm not going to pay for three coding copilots. We're either going to use cursor, or we're going to use GitHub, Copilot, or we're going to use one of these things and there's going to be a bunch of these markets where people are going to be like, we're not going to pay for 10 of these things. We're going to pick one. And so I think we're in a moment where just the budgets are big. And I think no CEO wants to go to the board and say, we're not heavily invested in figuring out how AI is going to impact our business. At some point, they're going to have the answer. They're going to make. AI right now impacts this part of the business. So we're going to spend money here. We tried this thing. It didn't work. And we're not going to continue to spend on that. We might revisit in two years and it's going to settle up. But right now it does feel like a gold rush.

How would you suggest founders who are building an AI communicate their traction and revenue?

I'm very empathetic in that. I talked to one of our founders who competes with one of these companies and I told him, I go, what do you think about that hockey stick growth? He goes, they probably have massive churn and go I probably agree with you. But I also know that that's the yardstick that we're going to get measured by when we go out to fundraise because that narrative's out there. So we too need to make sure that we're on a very aggressive growth ramp when we go out. I think it's hard for founders because on the one hand you're like what do those numbers mean? The real is that revenue going to be retained but it's an anchor point in the mind of the investor and if you're at a million of ARR and someone else is at 10, they're going to go, well you've had the same amount of time and resources, they're at 10, you're at 1. Why wouldn't I pick the team at 10?

It's almost asking investors actually to be a little bit more methodical about the way they're thinking about it because what they deem to be as the benchmark become the benchmark and leak out into founderland and then that kind of creates the churn of trying to embellish out the traction that you have and see it so future looking when it's impossible to do so just given and curiosity.

Revenue and half the time part of the story that is leaked is the part of the story that is most favorable. When someone tells me they had 200% NRR, their net revenue retention, that's really impressive. That tells me much more about the stickiness and utility of your product than a top line revenue growth number where I don't know anything about usage. I was at an event, I talked to someone who has bought, I don't want to name the product who bought a big AI product and they sold it top down and so the C suite signed off on it and I think they paid $2 million for it. And I'm like, well what's the usage like? He's like, nobody uses it. I'm like what do you mean he? Well it's a really cool tool but everybody in our org either has like a chatgpt or cloud subscription or has found some other collection of tools. So when management bought this they were like super excited and everyone else is like if you'd ask us, we would have told you we've already kind of mostly solved that problem. And so on the books that company's got $2 million in AR, but the odds that that contract gets renewed, very low.

So fascinating. I know we talked about it a couple podcasts ago, but you hear in this industry the high highs and then the extreme lows, but nothing typically in the middle of just like the companies and the founders who are just showing up day to day and just grinding. I believe all founders are doing that in good faith, but those stories don't typically get shared. So I understand why there's this feeling of wanting to over embellish when you can. It's not like it's fake revenue entirely.

And there's all these other weird incentives. Like I saw someone oh, one of the things we're looking for now is ARR per full time employee. I'm like well so companies are just going to hire contractors. They're just going to hire contractors and they're going to have fewer FTEs. And this is the eternal battle. VCs pick a metric that has meaning and then once the metric is understood, it is gamed, losing most of the meaning that it once had. Like as soon as NRR became a thing, I met companies that were intentionally selling in small contracts so that they could have account growth, very rapid account expansion. Like if you just look at the total size of the deal 12 months in and don't worry about how much it grew over time, you would actually get to the same answer. But I think founders are clever and the window to exploit quotes. As a vc, these metric inefficiencies is getting smaller and founders are smart and they're figuring out if that's how I'm going to be evaluated. We are going to make sure that our data when presented, puts us in the best light.

Moving on to our last topic, Lawrence Lundy Bryan released a new post from his some stack state of the Future called Data Driven VC is Over. In the post he makes a pretty bold claim. Venture capital's reliance on data driven research is becoming obsolete and the next era of investing will be shaped by intuition, contrarian thinking and human wisdom. The argument is that AI powered deep research research has made market research, thesis development and diligence instantly accessible, removing the traditional edge that larger firms had. When everyone has access to the same research and insights, investment strategies converge and firms all start chasing the same deals. Lawrence predicts that the best opportunities won't come from obvious trends that AI can map out. They'll come from decisions that defy data driven consensus. Companies that maybe didn't make so much sense on paper, but succeeded because because of just deeper intuition. The key takeaway was that this pattern recognition, intuition and contrarian conviction will become the new differentiator. And so in this world, solo gps, emerging managers with deep sector expertise and maybe stronger founder relationships will have an edge over mega funds bogged down by institutional processes. I'm actually curious to know at the GP level what sort of chatter is occurring right now with more folks leveraging Deep research across their funds. I'm sure that some are like this is a game changer and others are, this is just as good as what I have my 70 person team pumping out. Do you agree that it's leveling the playing field for market analysis, making traditional information advantages irrelevant?

So I use deep research a lot, not just for work. I found deep research allows me to get 70% of the way up the curve on an unknown topic. If there's a fairly robust amount of public data available, which before that would have probably kept you out of the category, you'd have been like this new industry, who are the big players? How do I learn about it? You'd have to go off and do research or you'd have to have a team of associates doing these relentless marketing maps, finding every company. And I do think there was a period of time in which simply seeing more companies and seeing more data was an edge. But now I assume that every big multi stage firm has. Well, I know they all have teams of engineers that have built data analytics capabilities that are sucking in all the data they're getting from both their own portfolio companies and from companies that they've met in diligence. And then they publish these charts. Best in class NRR for Series B companies is 285%. And you're like, where did that come from? Well, that firm is a growth stage firm that met with a bunch of companies and says we know what distribution looks like. So I think the thing that deep research doesn't yet solve for is one of the benefits of investing in private companies is that the information private and is not easily accessible by AI yet. So I still think there is an information advantage. If you are seeing more companies at your stage and you have the ability, I think more it just means I'm much more willing to invest in categories where I didn't know as much because I can start a cup of coffee, fire off a deep research query and get back clues on where to look if I want to learn more. And I do think it's going to take away some of the sort of tedious markets, market mapping and background work that analysts and associates used to have to do. But I also think that discovery process is a skill and I think if people just say I'm just going to assume that whatever deep research tells me is true and complete, they're going to miss out on stuff. And as someone who doesn't do a lot of data driven investing, I of course was the most resonated with me.

It will be interesting to see also, just because for a lot of funds those materials are seen as collateral that can push them into new circles or hopefully win founders over. So it'll be interesting if you there's almost a democratization of access to this what advancements you end up getting and is there just a clear diminishing point of returns for that effort? Is there signaling if you go into deep research and there's so much information on this market you're looking into, then maybe it's just not quite the area that you want to be spending a lot of time in. As an early, early stage investor, the article suggests that the best investments will come from decisions that feel irrational based on available data. The kind of kinds of bets that bigger funds might struggle to make just given the deeper processes that exist there. I do wonder that bigger funds because they simply have more money that they need to move. Is this an argument that they maybe might have the luxury to make such bets in a haphazard sort of motion?

Well, there's multiple dimensions of data driven investing. One is data about the company KPIs, the other one is data about the people. And I do think the big multi stage firms have gotten very good. I don't mean this in a bad way of algorithmic investing around people, people based on where they worked, how many companies they started before. If you said hey, who are the first hundred engineers at OpenAI, I bet you every big sanhub firm has their names, probably their birthdays, probably have them in Harmonic, maybe put an airtag in their back. There is a lot of algorithmic investing I see around people. The impulse is like hey, this is a person in a very important role in a very important company at a very important time in that company's life. On average I think this person knows something. But we've been doing that forever in venture and the TLDR is like it doesn't always work. A lot of them. Most impressive, interesting companies I've seen are people who were not wildly successful before and who were not super early. If you just said, hey, I'm willing to invest in anybody who's one of the first hundred people at OpenAI, that's kind of all I need to know. You could run that play if you have access. And the reason I think the algorithmic stuff is interesting for the big funds is they have access to, generally speaking, their brands give them access to the best people, either because they're already on the cap table of those companies or because those founders will seek them out. I don't think that's the same for small funds. And look, I would argue that in the last 10 years we hit peak data driven investing because our entire industry basically shifted from a healthy balance from B2B to B2C to almost exclusively investing in software as a service models. In B2B, excluding a few brave people who did consumer or hard tech or deep tech, most of venture capital became the business of funding business software companies and comparing them on a pretty strict set of metrics. Churn growth rate, MRR revenue per FTP. There was a set we'd gotten to a place where everybody was kind of running the same playbook and the only difference was how strict were you going to hold the those companies to the metrics. And the point about data driven investing, the thing I admire most about some of my friends at these big firms is the ability to spot a company that feels expensive in the moment, but understand that relative to the terminal endpoint, a billion dollar valuation or I just think about like David Z doing Facebook at 550 and everybody thinking that was crazy. And it was crazy, it was crazy low in the grand scheme of things. People got into Stripe at a billion, people got into OpenAI at a billion. Still TBD, but the data would suggest that you're up 300x not counting for dilution. And I think that's a different of bravery is the ability to see a company that's already valuable and be able to envision a world in which it becomes significantly more valuable. And I think people like to knock the big funds that I'm like, I think it's very hard to see a company like Stripe and say I know it's at a billion today, the ceiling is a hundred. And you'd be like, well you're not going 100x from a billion unless you do.

I think that those are the outcomes that they need. So it makes sense for them to jump into those companies. Whereas I think of, I don't know, emerging managers were on fund one and fund two. Those swings and misses are very painful because you come in at such a high price. We're seeing this divergent margins and the split out in types of deals that folks are looking at just given how big their firms are. There are deals that come in at prices that larger multi stage firms are best positioned totally and should. So it really will be all interesting how it ends up. We are all out of time for today, so thank you, Charles, and thank you for listeners for joining us here live in person for the Learning Corner. Please like and subscribe and we will all talk to you next week. Thank you.

See ya.