Building Two Capital-Efficient AI Companies: Arijit Sengupta, Founder and CEO of Aible and BeyondCore (Part 3)

Arijit Sengupta: We had done a project with McKinsey where we looked at 30 million patients across a million variable combinations. At that time, McKinsey had come out and said, “What would have taken us three months, we can now do in two hours using the software.”

Clay ended up writing an article around that time in Harvard Business Review saying BeyondCore was disrupting the consulting market. It got us into a lot of trouble with consultants.

That’s why I had to switch to go to the end customers. That article let the cat out of the bag. That’s how you bridge it. You find what you need at that moment in time. If I hadn’t gone after the data and tried to go straight for the money, I couldn’t have gotten there. AI wouldn’t have learned anything. The customer would have no reason to have a differentiated offering.

Sramana Mitra: You were a little bit ahead of the market. In general, investors always ask entrepreneurs in the context of AI funding is, “Does it work?” Unless you have some use cases and real customer data, you cannot tell whether it works or not. AI is very difficult to fund without getting some of the proof points in place.

Arijit Sengupta: Can I just jump to Aible? I know we’re doing chronologically, but I don’t want to miss this point you just made. When we started Aible, one of the things we did is, we went to Berkeley during the AI Summit.

We ran a contest where we took a bunch of high school kids, history majors, and MBAs and put them up against expert data scientists. These were Masters students at Berkeley who had already been data scientists. After one hour of training, two of the high school kids beat every data scientist.

On average, users using Aible were twice as good as the average data scientist. Then the data scientist said, “This is unfair. Two hours is not enough. Give us two days.” We gave them five days. After five days, only four out of 11 data scientists beat the Aible users.

We could have ended up with egg on our faces. There was a possibility that the data scientists could have crushed the Aible users. What happens in the market today is, people do so much selling and overstating the case of what’s possible. We wanted to have this independent situation.

We could have gotten embarrassed, or we could have proven that our stuff works on an objective dataset that is publicly available. The reason VCs ask for proof is so many AI companies hype themselves up.

Sramana Mitra: Every time there’s a buzzword that’s popular, all the companies go and pitch that buzzword. Let’s go back to BeyondCore. You’ve got Menlo Ventures to fund you in 2014. 

Arijit Sengupta: 10 years after working on the problem, I got funding.

Sramana Mitra: How much were you doing revenue-wise?

Arijit Sengupta: It wasn’t a huge amount. We only had four or five people. We didn’t have that high of a revenue to be cash-flow positive.

Sramana Mitra: But VCs are not interested in cash flow; they’re interested in accelerated growth. If I were looking at your 10-year-old venture in 2014, I would ask how would this company go to $100 million in five years? What was the accelerator that was in view at that point?

Arijit Sengupta: Let me explain what BeyondCore did first. We automated analytics. In those days, people were creating various dashboards. BeyondCore would try many combinations.

Think of it as drawing a million graphs. It’s statistically evaluating every graph to figure out how well is the overall behavior explained by this graph.

Then it would turn it into a PowerPoint deck or a Word document where we’ll talk you through it saying, “Your revenue is doing well in the US, but not in Germany. The reason it’s doing badly in Germany is that this product is selling badly in Germany.”

This segment is part 3 in the series : Building Two Capital-Efficient AI Companies: Arijit Sengupta, Founder and CEO of Aible and BeyondCore
1 2 3 4 5 6 7