Thought Leaders in Artificial Intelligence: David Talby, CTO of John Snow Labs (Part 7)

David Talby: We have two customers and two different specialties who do that on top of what we do. Quite a few companies have been building on top of our library to sell to different markets.

We don’t intend to compete with our own customers. We intend to stick to that pass layer and focus on data science. There is a lot more work to be done on multi-model learning and understanding images. There is a lot more that we can do in the space. 

Sramana Mitra: What portion of your business is independent software vendors building on top of your product?

David Talby: I would say at least half. 

Sramana Mitra: There are a lot of companies building products on top of your platform then. 

David Talby: Yes. 

Sramana Mitra: That is very interesting. I love what you are doing. I love bootstrapped companies. I’m thrilled to see this. 

David Talby: I’d been involved with three VC-funded companies earlier, and we decided that this was not happening with this company of ours. There is a huge overhead in those VC-funded companies. You talk, meet, sell with the VCs. You also need to explain to them what is going on with the board and have meetings with them.

I think that the worst aspect of it is that you need to skew what your customers want to what your VCs want because the VCs are the ones that you are working for. I see a lot of VC-funded companies base their decisions on what the VC needs for this valuation.

It’s not about what the customer wants. It can be pricing decisions, going to market, and whether they do professional services or not. 

Sramana Mitra: I was going to say that I think your model of a professional services heavy company is imperative to what you are doing because it is a specialized service. VCs don’t like that model. 

David Talby: This is only one of the areas where we have to think of customers. A lot of what we learn comes from them. One claim that we make is, we work with five out of the top ten pharma companies. On large-scale problems, we put stuff in production.

As you can imagine, the first time we came there, we said, “We have this pre-trained model and it’s the best.” Of course, it didn’t work. It didn’t work for privacy reasons and biased reasons because the documents were scanned images with handwriting.

This included the jargon that did not include any terminology because the definition of simple things was different within the same company depending on the use case. That is the point. A lot of the things that are unique about our products are things that we have learned from our customers. We took that learning back and implemented it. 

Sramana Mitra: With the companies building products on top of your platform, how do you charge them? Is that a royalty fee model? 

David Talby: The software is licensed on a subscription basis. It’s by server or by cluster. As they go to production and they need more servers because they have more customers, then in a way that becomes part of the cost basis. 

Sramana Mitra: I really love what you are doing. I love this complex domain-specific AI project. Thank you for your time. 

This segment is part 7 in the series : Thought Leaders in Artificial Intelligence: David Talby, CTO of John Snow Labs
1 2 3 4 5 6 7