2.28-29.24-Ethics-catalan.ai

ISHAAN GROVER: Hi. Good morning, everyone. I'm Ishaan. I got my master's in machine learning from the MIT Media Lab back in 2018. And now I'm finishing up my PhD. But I'm also the cofounder of catalan.ai, which is a B2B, SaaS, AI, dynamic pricing as a service company. And I understand that's a little bit of a mouthful. So I'm going to walk you through what we do exactly in five minutes.

But before that, I have a quick question. How much do you think this banana costs? Just an approximate estimate. Costs about $0.15. But the real question is, Where did this $0.15 figure come from? How much does a sweater cost? Or why does a mic cost what it does?

We were curious. And we asked around, from the smallest to the largest businesses. Almost everyone said something along the following lines. Well, I'll do cost-plus pricing. So I make my product for $100. I'd like $15 profit. So $115.

Or I'm the market leader. I decide the price. Or I'm always going to be the cheapest in the market. But these are high-level rules. And the demand for each product actually depends on a lot of things.

So for example, day of the week, holidays. Is this holiday causing an increase in demand or decrease in demand? Payday, events, seasonality, and very importantly, where are my competitors pricing, right? But prices should be based on supply and demand. And the high-level rules that we just talked about don't really consider any of this.

Now, think about the inefficiency in pricing on one product. And multiply it by all the products that you're selling. And that's a large amount of money just left on the table because of inefficient pricing. So we at Catalan decided to solve this problem once and for all. And we use AI and machine learning to go about solving this.

So we take large amounts of data, and we curated specific industry-specific models for groceries. And I'll talk about all the other industries in a bit. But we created these models. And what you get as the output is something like this graph.

So on the x-axis, you can see the price. And on the y-axis, you get revenue. And you can see what your cost is. In this case it's, $20. And you can see where competitor one is setting that price, where competitor two is setting that price. And you know it's a holiday or maybe two days after a holiday.

And your job now is to set the right price for tomorrow such that you maximize your revenue or profits or whatever you care about. So what this model is telling you-- that this merchant was pricing at about $22.79. And they made about $270,000.

But if they had priced at $22.39, they would have made $370,000. And that's $100,000 right there by just changing the prices. So in a sense, what this model is telling you is what you would have made at different price points.

But we weren't just satisfied with creating models in our little workspace and call it a day. We wanted to run pilots. And we ran many, many, many pilots over the last one and a half years. And I want to focus on one of them.

And I've gone a little too far. But this is with a major oil and gas industry, a major retailer, one of the largest in Latin America. And what we did was we took 12 stations, divided up six and six. And these guys, these stations were moving parallelly before the start of the pilot.

We ran a six-month pilot where we changed prices on the billboard that you see outside the gas station for a whole six months using our machine learning models. And we generated a profit lift of 32.4% and a revenue lift of 17.2%.

But we didn't stop there. We then went into groceries and generated a revenue lift of about 15%. And then we ran a lot more pilots-- so fast fashion, jewelry, clothing, apparel, shoes, vitamin A and C, the kind of products you might find in a CVS. And in all of those e-commerce industries, using our machine learning models, we generated a 27% average profit lift.

So how do you work with us? Well, you get in touch with one of us. And we ingest your data. And the kind of data we want is what prices you set historically and what your demand was. And that's it.

We'll do all sorts of network optimization. You have peanut butter and jelly. You change the price of peanut. Jelly is now selling a lot. So we'll do all of that network optimization. And we'll tell you exactly what your price should be tomorrow or for the next one month, if that's what you want, every single day.

So if these problems resonate with you, please come chat with us at our booth. And if you're in one of these industries that we've explored, we'd be happy to chat. Or if you've not thought about another industry, please come talk to us. We'd love to chat. I'm Ishaan, and it was a pleasure meeting all of you. Thank you.