1Mby1M AI Investor Forum: Shripati Acharya, Managing Partner at Priven Advisors (Part 2)

Sramana Mitra: I think what you’re saying is, the entrepreneur team’s domain knowledge is really what you’re putting the premium on to build something that is intense in workflow or domain-related value add. The AI is an enabler of that value add but not the critical piece. The AI technology itself is not the big enabler.

Shripati Acharya: That’s correct. It is a very critical piece in the sense that the technology is very important, but it’s not all about the technology.

Sramana Mitra: Let’s do a couple of examples of what you have invested in that follows this general framework.

Shripati Acharya: We’ve very recently made these investments, and we haven’t announced the names yet, but I’ll talk about the companies and what they do. To illustrate the point, one of the companies fits into the manufacturing technology stack.

Currently, when you’re going through a manufacturing process, there are several steps in the assembly line. For example, you might be making a television set, or a dishwasher or an automobile. It goes through a series of steps. As a manufacturer, you want to minimize defects, because if a defect goes into the field, then you might have to do a recall, if it’s an automobile. If it’s a consumer good, you might have to probably repair it in the field.

The second thing is that it is essentially a false negative, right? The whole thing went through the line, but the system didn’t actually recognize a defect. There’s also a problem with the false positive. For instance, you actually flag a defect, which isn’t there. Then you might have to stop the assembly line, and of course that has a direct cost impact in terms of the number of goods which you are producing, and of course, all the overheads of restarting the line.

So, accurately identifying and acting on a defect has been an issue in manufacturing for a long time. The way it has been solved is that you put fairly special purpose hardware at certain points in the manufacturing line so that you can exactly get the angles and everything straight, and then act on it whenever the alert is flagged by the software.

With AI and the vision models, this thing can actually change in multiple different ways. Now, it’s possible to use very inexpensive hardware and very good software to dramatically increase the accuracy of defect detection. It can reduce both false negatives and false positives.

That’s one, but to do this kind of a solution, you need to understand how it fits into the manufacturing workflow.

Sramana Mitra: Of course.

Shripati Acharya: For example, in manufacturing, the network is not open. You don’t have web access from outside into the factory for obvious reasons like security implications or restricted access control phone. So, it must be an on-prem solution, for example.

In addition to that, the alerts and the user interface need to fit into the control systems, which they already use to run the manufacturing plant.

Third, you need to understand how this solution fits in, in terms of positioning it to the buyer. Who is the buyer of this technology? It might vary by the industry in which you’re. It not only requires deep understanding of manufacturing as a vertical but also the specific verticals within manufacturing. There are so many of them. Then, you’ve to position it in the right way within that and integrate with everything else around it, figuring out how we’ll support work, how will product enhancements work, and of course, pricing it appropriately.

So in this, the part of AI wherein you’re using vision models to do all this detection is of course very important, but unless it goes ahead and does the rest of the solution, it cannot be deployed or will be used by the manufacturers. For them, this becomes a very critical piece of software. So, that’s an example of a vertical software.

Sramana Mitra: Shripati, what is the background of the founders for this venture?

Shripati Acharya: In this case, the founders were working on this while they were still in school, essentially.

Sramana Mitra: So, first-time entrepreneurs, but they have interned in this area.

Shripati Acharya: Exactly. While in school, they’re working on it with their professors in designing the solution. By the time they were a couple of years out of college, this was the very first thing that they did. They’d already been working on it for several years.

The process was mainly about understanding the pain point of the customer. To your observation about having deep domain knowledge, you would have to meet 50 customers, be on the factory floor, understand how it is being used, what really matters to them, and then executing it.

This segment is part 2 in the series : 1Mby1M AI Investor Forum: Shripati Acharya, Managing Partner at Priven Advisors
1 2