10.25.23-Digital-Covariance.AI

MICHAEL FLEDER: Great to be here. I spent a long-- Mike Fleder, founder of Covariance, spent a long time at MIT, getting all of the degrees. And our research spun out into this company, which we do one thing, tell you about the competition.

So in the last 20 years, there's been a ton of progress on internal analytics, what's happening inside your business, the interactions that you have with your customers. But most of your customer's time is spent not interacting with your business. And what's happening across the street or at your competitor's has been a black box. And so the next 20 years are going to be about external analytics. Show me something that happens outside of my four walls. What happens when the customers leave my store? What's my competition doing today?

So what you see at a lot of companies today is folks buy a lot of market research reports, and staple it together, and hand it to management. And folks look at it skeptically, these research reports, and these external data sets, alternative data sets, third party data sets, they often disagree. They're not in the right units. And they're really hard to make operational decisions with.

And so what we find is folks often discard them, or distrust them, and just they're unusable for daily management decisions. Covariance is the trusted machine learning layer on top of all of the data you buy, all of your third party data, external data, alternative data, that's all the same thing, commercially available data combined with your first party data. There's a need to combine data sets into a trusted source to make daily operational decisions.

And we do this in a-- we call unified modeling single model of everything across all the data sources that you have, which is-- and the ability to do that is founded in our MIT research. So your view today, let's say your competitor's public. Your view today will look like this. You know your own numbers, what's happening inside your business. And your competitor will report one number every 90 days, less internationally.

And so what Covariance does is we fill in the blank. We show what's happening right now across the street. Or what are your-- what are your customers doing when they leave the store? Who's turning?

The technical challenge here has been, well, the last 20 years, there's been a lot of progress in big data. And so there are lots of inputs and potentially lots of outcomes to learn from, lots of images that have been labeled in ways, or lots of words to predict, for example, lots of outcomes. The problem in this area is that there are very few outcomes that you actually observe with ground truth.

So for example, you may rarely learn with certainty anything about your competition, their financials or their customers. And so you encounter not just a big data problem on the left of inputs but a small data problem on the right in terms of how many things you know for sure. And so this is actually a new category of machine learning. And that's why the research-- and why our business.

And so our breakthrough that is the ability to synthesize not just lots of data and lots of sources, but lots of types of information all into one model to show you what's happening outside your business. And so the work we did for a Fortune 100 retailer, as an example, is a multi-resolution drilldown. So this is a lot of the theme of what we do is this zoom and enhance.

So I want to understand not just aggregate financials. But I'm trying to understand what's happening in a store. Or actually in this case, the effect of remodeling a store and how that affected not just my own business, but my competitor's business. Who did we steal customers from? Or did we lose customers in that process? And what was the economic effect for all the parties involved?

And go down a level further, and who are these customers? Not with the goal of identifying individuals, but just an economic understanding of what changes are happening to both my business and my competitor's business every time we make decisions. So we're looking for pilots for with both B2C or companies and asset management. We work across data sets that both US and global. Great being here today. And look forward to your questions.