1Mby1M Virtual Accelerator AI Investor Forum: With Sailesh Ramakrishnan, Managing Partner at Rocketship.vc (Part 2)

Sramana Mitra: So could you double click on your AI algorithm of finding companies and give us a case study of a company that you found with that algorithm? What were the heuristics of how you found this and what were you looking for? How did the algorithm find what you were looking for? How did it go about developing that relationship and finally investing in that company?

Sailesh Ramakrishnan: Apna is a company that we found in India. They are a blue collar LinkedIn for India. They’ve an amazing set of founders. The main founder Nirmit actually spent a lot of time here in the US, went to Stanford, and saw the value of professional networks. He then realized that this was one of the biggest gaps in India, especially at the blue collar level. So, you’d actually go sit with some of these workers in various factories and try and understand why it is that they were not able to improve their lives. That’s his story, but we didn’t see all of that.

What we started seeing was, he had created an app simply to connect people. This one app suddenly started showing up on our screens on a daily basis. Our algorithms rank various lists of global companies. So this company out of India was very interesting. We started looking at what was going on with this company. We were able to both talk to the company as well as understand what they were doing mainly because we had this unique mix of understanding of what’s going on here in the US as well as understanding the need in India because of our Indian backgrounds.

So we connected with this company just before their series A, which was exactly where the company was experiencing a point of inflection. Until then it was still a little bit more experimental. Right around the time we connected with them, the network effects of a professional network started coming into play. So we started seeing a rapid growth, both in terms of the app downloads as well as the usage and the value that was being delivered. So we realized that we were onto something really exciting. So we made our first investment right around the series A, and our algorithms continue to follow them. We invested at the next two rounds as well. Fast forward to a year later, they reached a unicorn status where they are one of the largest professional blue collar networks in India.

A quick update on them. India has been going through a significant surge in growth. One of the key needs for the country is jobs and especially increasing the spread and accessibility of job opportunities across the large population. Apna is going to become an integral part of the Indian government’s vision to enabling jobs across the country.

Our algorithms found this fast growing company at exactly that right time where that point of inflection was happening.

Sramana Mitra: There are a few heuristics that I picked up from what you said. One is you are monitoring certain apps or your algorithm is finding apps that are starting to rise in adoption in various app stores and so forth. That must be one way the algorithm found this company.

Then on top of that are heuristics like the founder’s background and schools and so forth – the usual stuff. What other heuristics is the algorithm deploying to find, to locate promising companies?

Sailesh Ramakrishnan: At this point we have somewhere between, 50-100 different parameters that our algorithms track. Things that are easily visible, like app store downloads, usage, retention, and web visits are all several of the class of traction metrics that you can see online. Even simple things like newspaper articles, number of patents, publications, information about the founding team and other employees, and even simple things like the kinds of jobs that the company lists, gives us insights into what is going on in a company.

A company that is successful usually makes their first critical sales hire when something great is happening. In the early stages, typically it’s founder who does the sales, but when the company feels confident enough to hire their Head of Sales, that usually is a signal that now things are going well within the company. You can separate the vast majority of things that we look at into 50-100 different distinct parameters. They fall into four different categories.

One set is traction related. The second is about the founder, the founding team, and the employees. The third has to do with other investors and what their opinions are about the company – who are the earliest angel investors, for example, which is also a very good signal about the quality of the company.

The fourth is the value that the company is generating in terms of other people’s opinions. It could be reviews, ratings, or comments from employees on Glassdoor, for example. So it’s a lot about the public perception of the company.

So, a combination of all of these things are distilled by our algorithms to determine both that this is an interesting company and that this is the right time for us to approach this company.

Sramana Mitra: Interesting, very interesting. And what are the data sources? Of course, there is scraping of the internet, but are you going into proprietary data sources to do this kind of modeling?

Sailesh Ramakrishnan: Absolutely. In fact, when we started back in 2014, that was the key question. What was the data going to be and whether there was enough data with insights in order to create or build on this approach? So, we started off by scraping the web to see what data was available. Luckily, we had thought about this idea at that time when a lot more formal data providers were starting up as well. So analytics providers, for example, regarding app downloads and web visits were coming online and that data became accessible. There were also other providers offering information regarding press clippings and newspaper articles.

Over time, we have sort of an evenly balanced combination of information that is available both from a structural perspective as well as what we crawl. Our most recent data set happened to be from the financial interaction side. We’ve been consuming credit card data, at a very summarized level as well. So, things that you may not even think of as a signal end up becoming a signal.

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