The Exchanges

Q Great framework. Thank you. And a little bit to the last point you made. You hear some investors talk about, um, horizontal AI versus vertical AI. So horizontal being enabling AI that targets broad use cases, vertical being industry specific. Does that matter to you, or would you invest in a horizontal AI company?

A Yeah, it matters a lot. Um, you, you said this really early on, and from the very beginning of our fund, we just said we're not going to invest in anything that's horizontal. Um, And the reason is fundamental, right? Like, if you understand how to build a machine learning model, all the fun is in, like, tuning it for its specific purpose. Um, and not all the fun, all the value is in doing that at an algorithmic level, um, but also at the data gathering level. And we, so that's the first thing we thought, um, To, to make us only invest in vertically focused applications, because that's where you can really get ahead of everyone else by focusing on tuning a model for a very specific purpose, getting data to train a model for a very specific purpose. Um, so that was on the one hand, like a fundamental understanding of how this stuff works made us think you have to be vertical. The, on the other hand, uh, it's very clear that with this huge shift to cloud, as we just saw, like we're still only halfway through this shift, All these massive companies that are the cloud utilities, I call them, ah, or cloud infrastructure providers, they want to get all these machine learning work, machine learning workloads right onto the cloud because they're really data and compute intensive. So they have a huge incentive to give out whatever tools they can for free to get these workloads into their…

AI assessment note: “we just said we're not going to invest in anything that's horizontal.”

Q Do you want to give a few examples from your investments of companies that have a unique or interesting data asset?

A Yeah, for sure. I'll start with a, something that's pretty simple and you would have thought is already done, but, uh, but hasn't been. It's a company called Constructor, and they give you a search, search box. That you can, you can put on your website, right? If you've got a media site, e-commerce site, whatever else. Now, Algolia, Elastic, like, all these companies do that. They make it very easy to deploy a very fast search box on your website, but the thing is, because they guarantee you that they're not going to share any data in any way with anyone, um, they can't really improve that search function over time in terms of, like, the autosuggest results, um, the ranking, Um, of the results. Once, once they're sort of shown up, showed on a page. Constructor pulls data across all of its customers. So it's got a bunch of e-commerce customers, media sites, and massive ones as well, like Jet.com is, is a customer of theirs. And it pulls all this data about what people are searching for, what typos they make, um, what time of day. If they're searching on mobile, do they have shorter, longer searches? What does that mean? What are they trying to find? Use all this data to figure out, like, what people are trying to do. And then look at all the click-through rates on all these sites, and, and provide a really, really accurate self-learning search engine. Um, you know, a lot of the,…

AI assessment note: “Constructor pulls data across all of its customers.”

Q So, um, you've been investing in those companies for a little while now. Any lessons learned in terms of what works, but actually what doesn't work?

A Yeah, um, just going off that example actually is, it, it's a good follow-on question, because our main lesson is, you've got to get customers to buy into the data network on day one, and if they don't, you should reject them as a customer, and so what I mean by that is, you have your terms, and your terms are, if you use our product, we will aggregate the data, and we will build models on top of that data, and we'll share the results or the improvements That improved model will be used across all of our customers. We're not going to share your specific customer data with another competitor of yours, but your data is going into a pool, and that pool makes the model better. So they're in your terms from day one. And your customers should either sign up to that or not. And because a lot of customers will say, well, that's scary to us. We don't want anyone seeing our data. And that's not exactly what's happening, of course. It's aggregated, anonymized, whatever else. But they might be scared by that. And some of those customers will ask you to do an on-prem private deployment, will ask you to not use their data to train anything that you do, and you need to reject those customers. Um, because you are not going to build a company that has any sort of moat around it, that has the world's best model, unless you get their data to do that. Um, and this is sort of like, ah, you know, 10…

AI assessment note: “our main lesson is, you've got to get customers to buy into the data network”

Q and then we'll open it up to people. Um, so I guess what's next? Where are we in the cycle of, of AI? This, this word saying where, because it's been so hyped up, um, some people are starting to say it's almost already over. Other people are feeling that it's actually just getting started, and it's a trend for the next 10 years. Where do you think we are?

A Yeah, I think we're at a really interesting point. So if we go back to that curve, um, I explained before, you know, with risk and time, and you've got the consumer stuff, you've got the AI enhanced sprinkling of AI stuff, you've got the AI centric stuff, and then you've got the AI enabled stuff. We're really at the start of this AI enabled age. So that is doing things with AI that humans just cannot do. And that is things that are involved or solving problems in very, very complex systems. So they are things like, um, Energy systems, logistics systems, healthcare, and food production, um, working with a lot of biological systems, whether it's food and agriculture, or whether it's human biologics. That's where we are, where we're starting to see AI be very useful there, and if we think about, I'd like to sort of frame that in two ways. Like, what, why are we starting to think about that right now? Well, one, the technology's getting good enough, and it's not that, you know, um, Deep learning does keep getting better and better. It's more that we're getting better at putting different methods together, right? So putting Bayesian methods in with deep learning methods in with, there's, there's a bit more attention now on probabilistic programming as a, as its own separate thing. So we're starting to be able to have an ensemble of models that can, that are good at different things,…

AI assessment note: “We're really at the start of this AI enabled age.”

Q How deep do you go precisely into the science? Do you have, ah, do you have, ah, partners or people you work with that actually, ah, look at the technology itself in depth?

A Yeah, we do it ourselves. Um, and again, we're less focused on, you know, can you execute to build all the features of the product that's gonna be built on top of this AI engine? That's not something that we view as super risky, like, the sort of caliber of teams we're talking through, talking to, they're gonna be able to do that. Um, we ourselves just walk through the experiments. Um, and yeah, we all have a background in the space, all three of the partners. Um, and so we're able to do that at some level of, of, uh, uh, fluency. Um, but it's, it's, it's very simple, right? Like the, You just ask a set of good, um, scientific process questions, um, and you, you find out really quickly.

AI assessment note: “Yeah, we do it ourselves.”

Q Okay. Very good. All right. So let's jump right in. So, um, what's, what's interesting to you guys Um, so the world of analytics, machine learning AI, that's, uh, sort of narrow, but at the same time very wide. What, what are some of the key, um, Yeah. Areas you find interesting?

A I might, um, I might sort of explain a framework we use, rather than sort of jumping into, we're interested in IoT security, which we are, and have just invested in that space, or whatever else, industrial analytics. The way we think about it is in terms of adoption risk, so if you think of, On the x-axis is time, and on the y-axis is risk. Um, it's really important to understand at what point is the state of the art in machine learning such that people can trust it to do the thing they want it to do. Because machine learning is inherently probabilistic, right, so it'll get it wrong some of the time. And that some of the time can be significant or insignificant depending on what you're trying to get it to do. And if you're trying to get it to diagnose something, ah, if it's wrong, Any of the time someone dies, and so that risk is very high. If you're getting it to put a recommendation up for, to buy something, if you get it wrong, a lot of the time it doesn't really matter because the downside is it's, it's a funny recommendation, or it's like an irrelevant recommendation. There's only upside. It's a very sort of convex payoff. So, to bring that down to where we focus today, I mean, you think about where we first applied AI. It was really in all the consumer domains where very, very low risk and only payoff. Again, a very convex payoff. And so it was like product recommendation…

AI assessment note: “I might sort of explain a framework we use, rather than sort of jumping into”