Episode #10: Euclid Insights, Benedict Evans

The question is, is ChatGPT a good business? And I suspect it is, in the sense that consuming $20 worth of compute for the average person, they're probably okay with it. But I also wonder, is this also like Uber where Uber was great when I could get anywhere in San Francisco for $5 because my Uber ride was subsidized by uber and the VCs who backed the company ultimately turned out to be pretty good because it created a transportation habit that didn't exist before, but it took private markets financing a lot of that to get through that curve.

It's the week of November 5th. Welcome to the Learning Corner, a Precursor Ventures podcast.

Welcome to the Learning Corner by Precursor and Happy Thanksgiving Week. I'm Mia Farnam and I'm joined by our GP and Managing Partner Charles Hudson, where we'll be talking through the articles and goodreads that caught our team's attention over the past week. Our first topic is from Euclid Insights. Titled the Siren Song of Services, this piece delves into into the concept of service as a software and how it differs from the traditional software AS a service SaaS model. Unlike SaaS, which focuses on delivering scalable self serve software products with reoccurring revenue streams, service as a software involves offering services powered by technology but far more reliant on human operations. These services often reach a larger TAM with highly personalized services, but frequently lack the scalability and margins that true SaaS models have traditionally benefited from. Euclid warns that the allure of service as a software stems from its ability to generate early cash flow and foster relationships. However, this approach can lead startups down a path where they are effectively building a consulting business rather than scalable product. As the article sharply puts it, service as a software is the comforting lie that you can build a software company while running a service based business. The piece highlights a key tension While services can help validate product market fit and generate revenue in the early days, they often create operational complexity, detract from core product focus, and make it harder to achieve the scalability VCs look for. Seems like in this time of AI services as a software companies are perhaps excited to win back some margin through embedding capabilities across the platform. Euclid states that companies should still tread lightly here since we haven't really quite worked out the extent and depth of capabilities. They gave the example of Klarna claiming that they can reduce two thirds of customer support costs through a partnership with OpenAI. I really doubt it, but wanted to know Charles, what's your view on services as a software compared to traditional software as a service models. And I'd love to hear that perspective, especially as it relates to learnings from the precursor portfolio.

Yeah, I mean, the Euclid folks making another appearance. I enjoy reading their writing. I think this all goes back to Sarah Benchmark's original post about service as a software. And in some ways I think the argument makes a lot of sense. And in some ways I think it's a reminder to how good a business model software as a service has been for venture capitalists. I think many of us are loath to give up this high margin, recurring, easy to model, easy to forecast revenue stream because it's maybe one of the greatest business models in the history of time. Sometimes it's hard to give up things that you love. When I think about it, though, I keep coming back to a couple of questions. One is, as you noted, SaaS was a great business model because it was recurring, easy to forecast and high margin. And if I think about this service as a software, one of my concerns is that we're trying to, like, bridge from the known and safe to the unknown and safe using a familiar model as opposed to, like, asking what are other ways that this could play out. So I think a lot of people are like, oh, we're going to sell outcomes, we're going to sell the work. And I don't know, Mia, like, one of the things I've learned from working with folks on the customer side is it's nice to be able to plan. It's nice to be able to forecast what you're going to spend on a piece of software. And sort of paying for outcomes, I think is very different than paying for seats. Like, seats have all kinds of other issues and challenges. But one thing that's nice about it is your CFO can say, all right, we bought 10 seats of Salesforce. I know exactly what those are going to cost for the year. And at the end of the year, we can ask ourselves, was whatever we spent on those seats worth it? I think with selling the outcome and selling the work, you don't always know what you're going to end up spending. So that's one. And the second thing, and this is where I keep getting hung up. I've met a bunch of people who are like, we're an AI SDR company. I'm like, that's great. People are like, well, these Companies pay these SDRs 60, $80,000 a year to do prospecting. Our solution is going to eat into that spend. I'm like, but you're probably not going to replace $80,000 worth of human activity with $80,000 worth of AI generated SDR activity. That's just unlikely to happen. So the real question is like, well, when you replace the human, what do you, like, what do you replace it with? And do you replace $80,000 of human with 50,000 of AI spend, 20,000 of AI spend, 10,000 of AI spend. Like, what's the number? And the thing I keep worrying about is if you have 50 people trying to replace an $80,000 SDR using similar but not the same underlying tech stack prompts, et cetera, that sounds like a recipe for dramatic price deflation. And it sounds like a recipe for a race to the bottom if you're not careful. And eventually maybe you end up replacing an $80,000 human with $15,000 worth of software that costs the service provider $10,000 to deliver. So now you've got certainly have a service as a software business that sells outcomes. But, but the cost of those outcomes and the revenue you make, it's like a worse business. The top line is lower and the margins are worse. But maybe you make it up for it by not needing as many people to build it. And if you can scale it across enough customers, et cetera, et cetera, et cetera. This is the thing I keep coming back to, is the reason most people would rather sell software over services is software has a marginal cost of 0 and it has all these great attributes I won't get into here. And services really don't have that. If you think about services as software, it's kind of an inertial way of saying services enabled software. And we've seen those businesses in the past, pre AI and pre AI, they're okay businesses, but they're not nearly as good as pure software, which is why so many VCs push companies to be 100% software and push down services, even though you can't really do that. So I remain really curious to see, as we replace humans or human labor or other activities with agents or AI systems, what does the margin profile look like for these businesses? And what effect does competition have in eating away at those margins?

Any conversations that are happening within our own portfolio where maybe the original business model was truly SaaS, and then now with the emergence of AI, now maybe founders are like, well, now we can kind of scale up this really cool enablement that can be highly personalized for a customer, and we can do it and win back some margins through X, Y and Z.

Yes, I Think the two interesting conversations I've been having with folks are like, hey, we figured out a way to use AI for our internal processes to make ourselves more efficient. Should we pass the savings on to the customer in the form of a better price, or should we capture these savings in the name of margin? And depending on the company and depending on the competitive dynamic and depending on their customer type? I've said the answer varies. It really depends. The second thing I've seen is a lot of AI native or AI first vendors have come into market to compete with companies that we have that provide traditional SaaS. And in some cases those companies have made bold claims. We're going to give you 10x improvement over what you have. And they're in. Our companies have a wow. If it's really 10x, we can't match that with what we have. And I'm like, all right, well, let's see what the customer actually gets. And in many cases the customer is not getting anything remotely close to what's promised, but they are getting something significant. And the question is, if you're at 2x, what happens faster? Does the AI native company that promised 10x go from 2x to something more material sooner than the company that's already got? The customer is able to integrate AI and sort of come down? I think the argument in favor of the new companies is it's hard to cannibalize a business you already have if you're selling software. I think the other thing is though, if you're the startup, you're kind of gated by how quickly the AI models get good enough to allow you to improve your process. And I'm actually not sure where those two curves intersect.

From your perspective. What advice would you maybe give to early stage founders who are trying to strike the right balance between leveraging services and staying focused on something that's more scalable from a product development stage. I felt like I've had a lot of conversations with awesome founders who have really good product and then you kind of look at the back end and you realize, oh, it's because they're running a dev shop. So no wonder their product is so good. But how can you possibly scale that at the burn that you're running it?

I couldn't agree with you more. I think services are a great way to get to know your customer. I love how we've like essentially rebranded pilots as design partners. You know, the nomenclature changes. But I think this notion that sometimes solving the customer's problem involves a combination of a software product and A service. And for me the question's always like, okay, does offering them services give you insight into things you can solve and product in the future, even if you can't solve them today? And if you think that providing services is a meaningful way to get insight into the customer problem and make your product better, by all means do it. And ideally you'll get paid for it. But in some cases you're just like, wow, the services component isn't going away. It's not something we can automate or make work. It's just something we have to do. I just think about, for example, there were a lot of companies before that would give you kind of white glove onboarding, which was a fancy way to rebrand services and like, hey, you give us all your messy data and we will make it clean and get into the product for you. That was a service that was unlikely to get materially better over time because everybody's data was messy. We might be in a world now where you can do white glove onboarding using AI and actually get better in your product at onboarding people's data. So I don't think we should be as afraid of services as we were before. But also I think it requires being honest about, hey, are you really building a product that where the services are informing what to build versus just a part of the product delivery that you're going to have to have forever.

Our next topic is the presentation created by Benedict Evans titled AI Eats the World, which is his annual presentation exploring macro and strategic trends across the tech industry. If you haven't opened up his presentation yourself, definitely take some time to do so. We've do our best to summarize a takeaway always on the pod, but there's something really entertaining about seeing it in visual form in slide format. So Benedict walks through the surge in interest and use case for AI across nearly every industry, while calling out the fact that we are still working out the questions about the space itself. One, how far would this scale? Two, how is this useful? And three, how do we deploy this? And while CAPEX is surging across the largest foundational LLMs and capital floods to the market to seize the opportunity, Benedict says, quote, everything is still changing under our feet. So Benedict calls out the fact that the consumer Internet model was to launch free, go viral and work out revenue later. That clearly does not work with the LLM cost model. OpenAI expects a loss of 5 billion on 3.7 billion in revenue. Computing costs alone will reach around $6 billion this year, according to a report by Foundational capital. So they definitely didn't launch for free. And on the morality piece, we in our VC bubble can believe things like ChatGPT have gone viral. While there's huge interest in generative AI, there's actually limited use so far if you look at the percentage of people who use it on a daily access point across the globe. So how long are we going to live in this gap of spending time, money and resources building and still not creating something that is exactly what the consumer needs?

Boy, I have so much to say. I always enjoy Benedict's presentations. I think the visual is really helpful. And there's some slides in there where I like, literally chuckled out loud when I clicked on them. They were just with that right level of like pithy humor. I think part of the challenge is everybody's struggling for an analogy for AI. Is it like Internet 1.0? Is it like SaaS? Is it like electricity? And I'm like, I don't know. I don't know if there's an analogy because it is certainly feels to me like revolutionary technology on the scale of like electricity. But the rate at which the technology is compounding feels. I wasn't around when electricity was. I mean, I'm old, I'm not that old. I wasn't around when electricity was invented. So I don't actually know what it felt like when electricity was invented. My sense is AI is compounding faster than electricity. And also AI is going to touch, even compared to the Internet. I have a feeling that's going to touch more elements of life faster. But I think if you settle on an analogy, it kind of dictates what you think is going to happen next. So if you're like, oh, AI is like fiber. It's like the fiber build out for the Internet. We laid all this fiber and it felt like it was ruinous and who's going to use it? And lo and behold, we used it all. And maybe it wasn't all efficient capex in the moment, but it enabled stuff in the future that was highly valuable. Or is it like the consumer Internet, where like, it is about being one of the early winners and building a brand. And I don't know what the right analogy is, but what I see a lot of people doing is trying to fit their expectations for where AI will go to some previous narrative. Well, like, what if it's just its own thing? What if it's just really different and there isn't a really good historical narrative or proxy? There are a couple of things I think about, though. ChatGPT at 20 bucks a month has kind of set the I don't think it's an accident that ChatGPT quad perplexity. And just about any consumer AI product you use is 20 bucks a month. I think it would be pretty hard to charge 50 bucks a month for a consumer product now because the reference price is 20. For me, the question is like, is ChatGPT a good business? And I suspect it is, in the sense that consuming $20 worth of compute for the average person, unless they're really using ChatGPT a lot, they're probably okay with it. There's probably an argument that there should be a ChatGPT Lite for five or ten bucks. But I also wonder, like, is this also like Uber where Uber was great when I could get anywhere in San Francisco for $5 because my Uber ride was subsidized by uber and the VCs who backed the company. And look, that ultimately turned out to be pretty good because it created a transportation habit that didn't ex but it took private markets financing a lot of that to get through that curve. Or is it like pets.com and you're shipping bags of dog food through a pretty brittle supply chain at a price that's contribution margin negative the second it leaves the warehouse. So for me, the real question is, so long as the big LLM providers have the support and faith of private capital markets, I don't want to say it doesn't matter, but they have time to figure all of this stuff out because they will be able to continue to foot the bill. And my belief in having talked to these folks is they all understand this flywheel, which is if they can make compute cheaper, people will use AI to do more stuff. And they can, if they get it right, drop the cost of compute faster than they drop the price to the end customer and still win. But that basically requires a deep faith in your ability to access, my guess is conservatively, another 10 to $30 billion in private capital for the next couple of years to continue to make these investments, which is kind of a scary value prop. That's a lot of money to raise. I happen to believe that OpenAI Anthropic and a few other companies will have access to it. The other thing is, I think customers are still trying to figure out what's all this stuff worth. I think right now a lot of people are like, sure, we'll spend a million dollars on a pilot to try a bunch of things with OpenAI or with Claude or with whatever model Llama 3, whatever we want to use. But I think ultimately at some point, maybe it's next year, maybe it's the year after the CFOs of these companies is going to come and say, we've spent of this money. What are we getting for this? Are we getting value from this? And right now I think it's the other end of the spectrum, which is like, we cannot be perceived by our investors or by the market as not taking AI seriously and not making investments. So we need to be deploying resources. But that pendulum at some point will switch and people will say, but we also need to be getting value from this. And I just don't think we're there yet. And when I read Benedict's deck, I was like, there's just so many open questions that he posed that I was like, I feel so much better about myself because I also have these questions and don't have an answer. But I think many people I know, their view and where I is headed is based on, I'd say a more confident view of where things are headed than I personally have.

I think it was his last year's presentation, which also was centered around AI. He had one slide that said, this is clever. So what a personal example. That's just funny. I, as the tech VC in my wedding planning era, I decided to cut down on questions being asked by my parents about what's going on with the wedding weekend. Created my own model within ChatGPT4O. And I pummeled it with months worth of communication, emails, documentation on every single thing that's going on in my wedding to pump out a model that I could pass over to my parents. And so if they have any questions, ask model, don't ask me. And I probably spent, I want to say, 15 plus hours pummeling it with every single thing feeding the model. You should have seen me when I had this done. I was like, this is literally the world's most amazing complete model. It's the going to knock everyone's socks off. It's going to be curing my problem. I mean, it goes to the gap in question of like, so why I pass it along to my parents. And I had one realization, one, I couldn't even share my 4O model because it had too many attachments and images there, so they couldn't even use the model. And then the second of which my sweet parents looked at it and they're like, oh my gosh, looks so great, I'm not using this. I'm going direct to you, like, what's the Point. And so I ended up just making a lot of screen record sessions of me asking the model fix for sending it to them, which they just thumbs up in my text message as opposed to actually absorbing it. So that goes back to my question of even on a consumer basis, the people who are using it for $20 a month, we still haven't quite hit it of this is clever. So what, what are we using for with folks who are debating spending millions of dollars a year on pilots, on contracts to use it to actually conduct business that they generate revenue on? We're, we're far from it, I feel.

Yeah, I read a great article on the information this week which was that, you know, a lot of the model providers are like, oh, maybe reasoning is the big unlock for business value. Maybe the pre reasoning versions of these models were kind of brute force. And I'm like, if I were in the model business I would certainly make that argument too. And having used O1, I think it's great. And so part of me feels like there probably won't be some big bang moment where we're like, AI is good enough, it's going to be incremental, incremental, incremental. And some people are like, it's, it's good enough, it's good enough. And I think right now I talk to people who are building things with agents and a lot of them are like, agents are getting better. I'm like getting better and good enough for production are actually two different standards. And so this is one of those cases where I think the people who are on the front lines building probably have the best perspective. But like Yubia, I've built some custom GPTs that I've deployed, some of them with more success than others. I will also just say it's important to continue to use these tools. I think about one tool where like my opinion has changed quite a bit in the last two months is actually perplexity. When I used Perplexity a few months ago I was like, just doesn't feel that different than Google in terms of the results I'm getting and the reasoning. And lately as I've played with it more, I'm like, oh, I am starting to get a feel for what's different. But again, at 20 bucks a month, paying for consumer search engine is a new behavior. And so I'm very curious to see what existing behaviors change in a world of AI and which ones stay the same. But 20 bucks for a consumer is a lot of money unless you're using ChatGPT to help you with your homework or some other use case. I don't know that Most consumers have $20 worth of queries and questions where ChatGPT or any LLM is well suited to provide context and answers above and beyond what they can get from traditional search.

Well, we are all out of time today. Thank you to our listeners. Please like and share with friends and we will all see you next week. Thank you. Bye bye.