categories

HOT TOPICS

Bootstrapping an AI Chatbot Startup with a Paycheck: Ankush Sabharwal, CEO of CoRover (Part 3)

Posted on Friday, Sep 1st 2023

Sramana Mitra: Going back to 2016, there were these platform providers but there were also these other startups that had full applications. Where was the gap? What is the justification of starting CoRover at that point?

Ankush Sabharwal: There were two sets of companies. One was the platform companies and the other is the implementation companies.

Sramana Mitra: There are three sets – platform providers, the application developers, and then the pure-play companies that focus on the same way you are focusing on that problem.

Ankush Sabharwal: The third kind is like us. That’s actually not the third kind. They had picked up open source and they were just building the nice UI and the nice flows. They were also more of the implementors than the actual technology developers. Underneath, they were using open source. The underlying framework was still open source.

Sramana Mitra: Is that you? Is CoRover using open source underneath? Core.ai and Yellow.ai are all open source underneath?

Ankush Sabharwal: Core.ai is a good company. I learned from them. I have less information on Core.ai. Other companies for sure.

Sramana Mitra: My point is there were other companies like you who were also were going after the same problem. We’ve done case studies like this. We’ve done Core.ai for example. You are starting out. What is your positioning? Are you going after B2C or B2B?

Ankush Sabharwal: Just before that, I’ll tell you what we did. We came up with the concept of Chatbot-as-a-Service. Few companies like Core have implementation partners who use our platform and then create virtual assistants. When we started, we said it was a Chatbot-as-a-Service. You give us contract. We have our own technology. We didn’t have to spend so much time in training.

Sramana Mitra: You provide the product and the service.

Ankush Sabharwal: Both. The tech was good so that we don’t have to give so many utterances in the classic NLP. Google Dialog Flow says you need to give at least 15 utterances per intent. We finetuned it. Even one utterance is good enough to provide 90% accuracy.

Sramana Mitra: What types of customers did you go after?

Ankush Sabharwal: We started with travel. We need more help when we’re away from home. When I’m in my city, I’ve got family and friends.

Sramana Mitra: Who was your first customer?

Ankush Sabharwal: Our first meaningful customer was IRCTC – Indian Railways.

Sramana Mitra: When you got Indian Railways, did you still have your job?

Ankush Sabharwal: In one of the companies I worked in, the founder offered to invest in my company so I could continue to work in his company. I tried for some time but said no. I needed to concentrate full-time. He actually gave me funds.

Sramana Mitra: How much did he invest?

Ankush Sabharwal: Only a small amount. He was not an investor. He said that I should just take it and give him the interest. When I paid, he said not to worry about the interest. He actually helped me with the amount.

Sramana Mitra: That’s wonderful.

Ankush Sabharwal: Still I could not manage my expenses. I joined my old company Altisource. From Altisource, I moved to Digital Hardware.

This segment is part 3 in the series : Bootstrapping an AI Chatbot Startup with a Paycheck: Ankush Sabharwal, CEO of CoRover
1 2 3 4 5 6

Hacker News
() Comments

Featured Videos