Building Two Capital-Efficient AI Companies: Arijit Sengupta, Founder and CEO of Aible and BeyondCore (Part 2)

Sramana Mitra: You started thinking about BeyondCore around 2004. What did that entail? Did you do this full-time?

Arijit Sengupta: Clay doesn’t admit this story as being true so I’ll tell you the story as I remember it. You incorporate the company as part of the business plan concept. I told Clay, “I’ve got this great job.” He looked at me and said, “All that business plan you did, was that real?” I said yes.

I ended up quitting the job, which was high-paying. I had some amount of savings from my Oracle days but didn’t have much money at all. I started BeyondCore and took on a part-time job.

As soon as I got my first customer, I ended up quitting the part-time job and went to BeyondCore full-time. That was still around 2005. We are talking very early days. The problem at that time was, unlike today, there wasn’t a cloud and all these platforms that you can build on top of. There was a huge problem that AI is only as good as the data in many cases.

There was a huge data quality problem with the data. Now, we’ve a lot of open source stuff to deal with that, but not at that time. I ended up doing a lot of fundamental research in terms of building AI that is resilient to bad data. How do you build an AI that can handle many different situations?

I got about 17 patents when I was at BeyondCore. I was lead inventor on all the 17. Around the 2013 timeframe, I got to a point where I felt, “This is working. We have the AI figured out.” We started getting real customer traction.

We went out to raise. We pitched to Menlo Ventures. On the same day, they gave us a term sheet. It just feels strange. Nobody would fund you. Then, you are ready, and the first time you pitch, you get a term sheet.

Sramana Mitra: Did you have customers at this point?

Arijit Sengupta: We did. We had several customers. We had gotten close to cash-flow positive, or we were just at cash-flow positive.

Sramana Mitra: Let’s talk a little bit about what you did to get these customers. Your point is absolutely well-taken. The real issue that we try to drill into all startups is, if you have customers and go to VCs with customers and product-market fit, you can get funded without much problem. Talk to me a bit about what problem were you solving for what kinds of customers. How did you acquire these customers?

Arijit Sengupta: There are two phases to this. To help your readers, I’ll go back and explain this part first. The first phase was, we needed a lot of data to train the AI. We were not getting paid at this point. We were just trying to get data.

We went to the 10 largest outsourcing firms. I had some connections out of Harvard Business School. Six or seven of the 10 largest started working with us. We would go to them and say, “Let us do some work with you to give you Six Sigma insights into your data quality.” We were not going to tell them, “We are going to do automated analytics.” That would be too highfalutin.

We were just doing something tactical that they valued. They were all doing Six Sigma evaluations on the quality of their employees. Now, there is a better way. We would make $10,000. Sometimes, we would make $5,000. We had it in the contract – anything the software learns, we get to keep. We don’t get to keep the data, but we get to keep the learnings.

That was phase one. Then we started going to customers. We’d go to large manufacturers and say, “Here is a real use case. Would you like to do this? You don’t have to pay us until we show you results?” Then we started doing this. Because we were doing it using the software, these projects would take us one to two days.

Customers were used to three-month long consulting projects. Willingness to pay was very high. We could find a price that was much below the price they were willing to pay, but it was profitable to us because we were doing it through software rather than manual consulting.

That is classic Clay Christensen. The job that was really profitable for us was really unprofitable at that price to our competitors.

This segment is part 2 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