Thought Leaders in Artificial Intelligence: Squirro CEO Dorian Selz (Part 2)

Dorian Selz: What differentiates us from the pack is that most of these companies predominantly use external datasets, for example, for public reporting of the company. Our specialty is that we effectively crossed the bridge from outside to the inside. Most of our customers predominantly use us to analyze company internal datasets. These are quality reports, email, and all types of meeting minutes.

We process these to extract signals and we do that in primarily two verticals. We do that in financial services where we have a pre-trained catalog of about 70 plus models. These include elements like intent to invest or intent to expand. There is an entire model catalog that we built there including debt, downsizing, divestment, equity raise, or bankruptcy risk. We also look at internal data. Then we apply that logic. You ingest data. You start to streamline data.

We think of that like a bit of an assembly line where you start to apply all types of little steps to effectively extract the insights out of those datasets. Once you have those insights, you need to contextualize, gather, understand, act. The key differentiator is being able to do that with internal datasets.

Sramana Mitra: This is something that I enjoy doing. AI and machine learning, in general, I really love working with. This problem is something that I’ve thought about. It gets my juices flowing when I encounter this problem. I’m going to take one example you gave of sales leads in the context of intent to invest.

We’re talking about a set of account managers inside of a financial services institution and they are talking to customers and managing portfolios. In that interaction, you are gathering signals for intent to invest. That’s what you’re saying?

Dorian Selz: Correct.

Sramana Mitra: What is an example of a data that triggers a signal like this?

Dorian Selz: For those generic models that we have trained like management change, that’s a potential way out of that business because new management wants to change everything. A management change for a good account executive is a moment where she will reach out to the other side to introduce herself to that new management. We have taken many press releases that these guys collect and redistribute.

We then start to identify what the concept of a management change is. That can be XYZ company has nominated John Doe as new CEO, or XYZ has promoted Mary Doe as the new CFO. There are many ways to describe management change. We have started to label all of these types of situations and create a machine learning classifier. I’ll be able to point to you that XYZ is one of your key accounts and this management is being talked about.

One of the most powerful classifiers is around market rumors. There is supposed to be a management change in that company. We essentially look at marketing reporting and pick up such rumors. The second dimension is the management change. If you have a cross-section of both like a Venn diagram, there’s high probability to identify management change and be ahead of the curve. Let’s say corporate banking. As a corporate banker, if you detect an intent to invest and intent to expand or divest, these are all elements where these global banks have multi-wallet situations.

Sramana Mitra: You have to win that deal.

Dorian Selz: It’s a competition for wallet share. We provide them with decision-making intelligence to be ahead of the curve to engage a customer when there is a high propensity that there is a deal about to happen. Whereas your industrial customers would need support in financing, trade financing, M&A financing, and expansion of credit line.

This segment is part 2 in the series : Thought Leaders in Artificial Intelligence: Squirro CEO Dorian Selz
1 2 3 4