1Mby1M Virtual Accelerator AI Investor Forum: With Sailesh Ramakrishnan, Managing Partner at Rocketship.vc (Part 3)

Sramana Mitra: Very interesting. All right, so now I’m gonna switch back to the more usual way I do the AI investment thesis discussion, which is to ask you about your firm’s perspective about AI investments and how are you thinking about it? What is your investment thesis in AI?

Sailesh Ramakrishnan: We are in the midst of a tremendous AI wave. We have been practitioners of AI through most of our careers. This is probably the fourth wave of AI excitement that we have seen in our own lives. I started in AI mainly because of this first or second wave that happened almost thirty plus years ago. So we have been seeing these waves, but at the same time, I think it’s important to note that this current wave is different. Every wave, unfortunately, has been oversold a bit, but this particular wave, I think is oversold, but I think underneath, a significant change has happened.

Sramana Mitra: It’s real.

Sailesh Ramakrishnan: It is. This one has a significant leap forward, much more so than the previous ones. There’re two things that clearly show that. One is the amount of investment from a wide, broad base – large companies and VCs like myself. Then the second thing is the pace of adoption. I think it is reasonably taken for granted that every company needs to have an answer as to how they are using AI in their organizations in order to at the very least stay in tune with the rest of the world. If not, they’re going to get left behind.

Now, going to the answer to the question you had asked about our thesis. So, we’ve started tracking the most recent wave since the significant values that come from our data-driven approach enable us to spot trends early. So in the mix of companies that are ranked by algorithms, the number of AI-related companies started increasing rapidly late 2018 onwards.

To give context, this was roughly two years before ChatGPT launched. We started seeing this excitement early on, and we started watching what they were trying to do and realized that there was a fundamental change. To set context on the timing, the transformer paper came out in 2017. 2019 is when we started seeing companies adopting those ideas and building things.

Our thesis started separating itself into four parts. First is that we realize that these models that are being created, these foundational models are going to be tremendously impactful, but are also going to be tremendously expensive to build. Therefore, while open AI, I think is an anomaly, where an individual company was able to generate this, we expected this fully to be in the realm of the large companies, the Googles and the Microsofts of the world. We realize that that area might not necessarily be a place of investment for a VC like us. We are a smaller size VC and hence, the amount of capital that was going to be required by those companies would be very large for us.

The second category, which was perhaps more interesting was that almost every one of these foundational models required a lot of tooling around them. Things like vector databases and execution automation to build monitors and evaluate the quality of these models required a lot of tooling. That became a very interesting area and we started focusing a lot more on it.

While these are going to be potentially successful companies, we felt that their valuation is still going to be capped as a smaller percentage of the total AI wave wherever the value was in products and verticals, where AI was going to get incorporated on a fundamental basis.

At the time we started tracking this, most of the applications were a little bit more than just technology demonstrators. You saw them applied to chat, or you saw them applied to images and videos. Applications weren’t arising yet.

So, we started monitoring actively for applications. We started seeing a bifurcation from our thesis. There were a certain set of applications that were literally low hanging fruit, uh, which were just applying AI or GenAI directly to a use case, but not with any significant advantage – no proprietary data, no deep insights into the needs of the product. It was just an efficiency game where you could replace humans with a reasonably good GenAI system.

Examples of this include things like copywriting or first tier of customer support being automated. So those were relatively simple use cases, which were good and important but would not lead to significantly large successes.

So, the core of our investment focus today is deep vertical applications of AI that either incorporate custom data that is not commoditized or not broadly available and an understanding of a deep customer problem or business problems. So that’s been our focus. That’s where we’ve been investing in.

This segment is part 3 in the series : 1Mby1M Virtual Accelerator AI Investor Forum: With Sailesh Ramakrishnan, Managing Partner at Rocketship.vc
1 2 3 4 5 6