Mike France (Proper):
We do two things basically, with our data set principally right now. We essentially maintain regularly a large set of e-commerce links for affiliate commerce. So that if you're a consumer and you want to buy a brand certain product at a certain store, we can take you right to that product detail page. Right to that listing page. We maintain that for our brand clients. They can put software on their website that essentially like virtualizes e-commerce through us put you click away from a store. We now have a direct cart where you can actually build a cart for us and check out through one of our partners in kind of a seamless manner. So we're kind of mimicking direct-to-consumer there. And then we give brand dashboard to understand all that data and their product distribution and stores.
But to do that, when we started this, we started, of course, with a few brands as a pilot set and everything done very strictly in combinations of regular expressions, right? And we're like, okay, this will definitely work and we know enough about the math here, but we clearly cannot scale this. And we also can't develop if we're just limited to every brand, we need to set up every brand. We can't create other products or services that leverage a wider data set. So with Watchful, number one helped us make the matching and classification of products for individual brands that we're matching a lot more robust. But we could also then do it at the scale of the entire data set. So rather than slowly eating the world a little bit and now we have now, our managed data set for our clients is like 15% market share of the entire cannabis market in the U.S. But with Watchful, we are classifying and identifying brands across the entire national set. So now we can utilize that to do a bunch of other things to create new products and services that we can then sell intelligence products we can sell to brands or retailers. Right? If I can't classify all 3 million recreational cannabis items every week or every month or how often would we need to do it, of course, then I'm limited to that. But once I can do that, I can start enabling store prospecting, store similarity and those kind of things. So number one Watchful allowed us to scale from what kind of started as let's say like 1% market share that were kind of managing in this way. To actually now for ⅔ of what we need to do to make this all work. We have complete national coverage for every cannabis item.
Shayan Mohanty (Watchful):
Mike France (Proper):
Classified brand extracted and then tracking those brand aliases. That's a huge metric right there essentially from like whatever you want to call it from like 10,000 to 3 million items classified regularly. The other thing here is just of course, if were to we'd have to back to the envelope, how long it would have taken us to do that. Otherwise but with Watchful, were you know, we are obviously we had a lot of subject matter expertise and we kind of had our numbers down very well we knew for strains, how are we gonna do that? We have thousands of strains, but they're very unique literals effectively. So like, that's kind of more of a data engineering thing. We don't need to do that. We knew where we needed the most firepower in particularly something like Watchful with classification. But it took us essentially a week to get that classification up and running. These were like I said, it was so intuitive because this is the way that we thought and understood things. But we had no way to get that knowledge into a system that could use it.
I mean, we actually talked about like, the reason this was found was we're like if I could just do probabilistic Regex-based waiting, right? And then of course, Watchful goes beyond that by then essentially automating that to I think you guys are essentially getting us out of the regs too, at this point. Yeah, exactly. You just need to now kind of put the most basic formation of your concept or whatever you're trying to do in the Watchful in a very simple manner. You don't need to be an advanced query writer. So again, I don't have an estimate for how many people would I need to be supporting a labeling operation. Or, we were gonna hire an advanced query person, right? someone who's an expert and Regex and now we're planning not to because we don't really need to because the reg-ex itself is kind of automated at this point. So what you really need is you just need some data entry, people who know what they're doing. And you know, we've cut back our hiring plans. And our overall kind of skill set need has been reduced because of this as well.