Sramana Mitra: The main question I am asking is at what point did you start productizing?
Nitesh Chawla: We started thinking about it from a services perspective. One thing that’s true today is that in the mid-market, you must have a partnership services model attached to it. It needs human expertise along the way. Having said that, in about 2013 or 2014, we were doing our first demo of Aunsight 1.0. In 2013, we launched it. We came out with a data platform.
It was still a platform that we were managing. The client was not directly using it. We were using it consistently in deploying solutions to our clients. We are a cloud-native platform. We connect to data sources, bring the data in, do the intelligent transformations on data, and get the right dataset out. Then the clients can use the dataset or the models that come out of the dataset. Think of it as a manufacturing pipeline that we were running and managing. It was ran like a manufacturing pipeline on a daily basis.
Sramana Mitra: You were going to market as a middleware platform company.
Nitesh Chawla: Yes. Every data science journey is a collaboration effort. There has to be this idea of collaboration built into our platform. Second was the secret sauce ultimately is the data model that I’ve created that is solving my banking or media problem. A couple of years ago, Aunalytics launched this product called Daybreak. Aunsight is a platform.
Daybreak is a product that we launched. The idea is that every day, I get the answers that I want. To get a set of answers, I need a data model that looks at all my databases, updates itself, and gets me an updated answer every day. I don’t have to call an IT person to run a new query for me. I have some cached answers I get. I can also ask questions in a natural language. You don’t have to know any SQL or worry about data refresh. The data gets automatically refreshed every night.
That has been transformative. What we are telling the industry is we have figured out which hundred you care about out of the thousand data fields. Within those fields, we also have smart features. Instead of running a model, we are putting intelligence in it where we are saying, “I’m going to make a prediction that this customer is likely to go.” The output of a row’s predictive model is a field in my database now. Any precomputations are put in the database as well. You have your answers without having to push a button. The banking folks are very excited about it.
Sramana Mitra: Here’s one piece that is missing for me. There is the plumbing of the data warehouse to the middleware. Let’s say the warehouse to the big data where you are doing the modeling. That is different from customer to customer and industry to industry. Where is that domain knowledge and the domain-specific business logic coming from? Is that something you are providing or are the customers providing that?
Nitesh Chawla: You’re correct. Every core system needs to have a different kind of connector. We have this product called Aunsight Golden Record. What that does is there are hundreds of connectors available that we have used or developed in service of a client. We have connectors for different core systems. That’s the first step.