2024 MIT Digital Technology and Strategy Conference: Lightning Talk - ProfitIsle

JOHN WASS: Hello, everyone. Thank you so much for having me today. So Profit Isle has its roots in MIT. My partner and founder, co-founder, Jonathan Burns, taught at MIT for about 30 years, and I'm also a graduate.

A lot of businesses right now are at an inflection point on AI, where the amount of data that they can get out of the public domain is pretty much reached saturation. And the real frontier is how to use enterprise data or unique data to just the enterprise itself, that private data. This is a huge opportunity. It comes up in the Google conferences, and this is one of the things that was written about here in Fortune or Fast Company.

This is exactly where Profit Isle exists, in the use of converting proprietary data to a usable data set so that corporations can use this information and ground themselves in their own operations. In particular, one of the major limitations to AI right now is the ability to use the proprietary enterprise data to actually drive AI models and also data governance trust and the ability to validate and not get lost in hallucinations on their own data. So all of these things are challenges that Profit Isle addresses.

So what's really exciting is it turns out that we do exactly what the keynote speaker talked about. We are creating synthetic data and we're doing it through the lens of the profit and loss statement, that tried and true core of any corporation. We show up with a corporation that might be using three or four profit and loss statements to run their whole business, and we'll create, in some instances, over a billion profit and loss statements.

We create a full PNL for every invoice line of every transaction, and we use their enterprise data to do that. And that transformation is a critical component of how we can actually use large language models to generate huge value for corporations.

And the way we do that is we can absorb all of the company's data, transform it into a value added PNL for every invoice line. And because of that, the grammar of our data set, when its output is in the language of the income statement. So it's easily absorbed, as any other language would be, in a large language model.

What we do is we've identified over the years 26 distinct data sets that are generated by corporations. We have a very clear way to absorb those into the profit and loss statement. So, for example, let's say that you're using GPS data to track your deliveries. We'll take that data and we will then calculate the exact cost of every delivery.

On Tuesday, you got stuck in traffic. On Wednesday, it took you 15 extra minutes to deliver. All that is included in the data sets that we absorb and creating that kind of visibility into what's really happening in the corporation. We do that across all the line items on a general ledger, and we take all the data that's listed here, which is very extensive.

That first transformation is one of the things that creates all these PNLs And this is what you, I now will call, basically, generative data. And the other thing that we do is we create segmentation data, which, again, augments the data. It creates even more value. The 80/20 rule is alive and well in every corporation that we've worked with.

By the way, we're approaching $600 billion of revenue through our model. So it's well vetted across multiple industries around the world.

In one instance, we're working with a $10 billion retailer. For them, we've created over a billion PNLs. We've taken in 13 data groups across one ERP system. And they're using it to monitor their promotional spending, understanding exactly which customers are profitable versus not, as well as what channels are profitable versus not.

We're using it to help them with store labor planning. With over 2000 stores, there are 2000 people making decisions about one of the largest expenses in a retailer. And that variability is something that AI can help with dramatically. And, finally, we help them understand the profitability of every product, every supplier, every planogram in every store. So that's an example of where we're using this transformation transform data.

Then we also have a global distributor. Again, lots of data groups, three ERP systems across multiple countries. Here, we're helping them see private label pricing, customer contracts, and product mix optimization. And we also work with global manufacture orders. Here, we're looking at multiple ERP systems across multiple countries, and we're helping them understand the actual impact of supply chain disruptions, variability in inbound raw materials, as well as customer negotiations.

So our ask is that we'd love to work with some of you in the audience. We work generally with very large multinational companies and, overall, we would love the opportunity to work with you, if you think there's an opportunity for you. Thank you.