#55: AI Solutions for Paperwork and Documentation
TranscriptionSantosh Sankar 0:40 Hey ladies and gents Welcome Back to the Future Supply Chain podcast. I'm your host Santosh Sankar. And joining me today: our very own James Coombs, CEO/Fo founder of Vector AI. Welcome!James Coombes 0:53 Thanks Santosh. Great to be here. Santosh Sankar 0:54 Yeah, so you're joining us from across the pond, and I would say You're a portfolio company that we talk a lot about when it comes to the industry. Equally, our listeners might not all be aware of what you're building. So we'd love to just kick it off with a quick, you know, two-minute overview of what Vector AI is and what you do.James Coombes 1:20 Yeah, sure. We often talk about the elevator pitch, right? So maybe you guys have total buildings, but for me, an elevator pitch goes like three or four floors. So we are Vector AI. We are we're a paperwork copilot, we help to automate paperwork and logistics space. And by paperwork, we mean all sorts of things from bills of lading, invoices, packing lists, and so on. We do about 20 different documents. 200 different data points across 20 different docs and growing, growing over time. And in terms of the scale of what we do, on aggregate. Our customers process about $130 million dollars every year at the moment, that's obviously growing. So it's a huge space and one that I think we can look obviously objectively or not objectively, I think we can make a big difference.Santosh Sankar 2:11 Tell us about the founding story behind Vector. Because you know, I remember first meeting you and Nisar back in 2017 on a trip to London in like this swanky or I guess I remembered as swanky it might not have actually been... coffee shop in Shoreditch. How did this even come to be?James Coombes 2:33 Yeah, so I don't know how swanky it was. I can't remember exactly the name of a hotel, but it was just a coffee shop. Every coffee shop in Shoreditch is pretty swanky I guess. We saw this massage. So I met my partner at an event called Entrepreneur First. And we met in Mid 2017 and we've been working together ever since. You and I met on some random Slack channel started by a guy called Sam Cropper. I believe his name was from In Motion. And I think you were on the chat as well. And it wasn't a very active chat. And I remember seeing your name and I remember you had basically said you are going to be in the UK at some point would love to meet interesting people. And I was like, yeah, you know, I think I'm interesting. I'll meet. So I dragged my partner along to this random meeting and just random coffee shop slash hotel review with a random dude. You. You put us in touch with a bunch of people literally from having just met. And I just remember being really impressed that you kind of go to those lengths to kind of work, work with us and help us out, given that you'd only just met us. And for us, that meant a lot and I obviously kept on, kept on going away and kept on. I kept the discussion up a review over weeks and months afterward.Santosh Sankar 4:24 You explored a lot of different ideas around the supply chain. What was that 'aha' moment where you realized we want to go be the co-pilot around paper and paper communication.James Coombes 4:38 I'm not sure if it was a moment specifically. I mean, my previous career work, I worked in a big bank. I worked in commodities and trade. A lot of my customers when I trade houses, they're doing a lot of trade finance. And so I was just very aware of all the kind of paperwork issues of around that business alone, I'd also set up had also set up a furniture company as well. At the same time, I was at some point in '08, ended up importing a lot of furniture from around the world into the US and Europe. And had so it was pretty, pretty aware of kind of all the moving parts around supply chain and freight forwarding as a result of that experience. So I felt like I had a pretty good overview of not just the kind of finance and the kind of trade and the trade finance world, but also just general operational aspects of the supply chain. And I think I've been able to combine those in a way that kind of gives us abasing an edge. And that that was how we kind of got onto this. And it was a problem and I think it's one of those spaces that is is a huge problem, but not that many people know about it, and that's exactly where I want to be as a company.Santosh Sankar 5:56 Mm hmm. I think the interesting thing our listeners might have picked up on is that the one thing you and the team at Vector hold to be true? Is that paper is a mainstay of the global supply chain? Explain that expand on that, because we obviously agree as investors that that's true.James Coombes 6:15 Yeah. Yeah. It's something we get, you know, occasional pushback on. Really what we're addressing is a standard problem run with supply chain and the very definition of a fragmented industry. I mean, you've got everything from mom and pop shops and kind of manufacturers around all around the world customs, houses, government agencies, banks, insurance companies, shippers, I mean, you've got so many different players in the space. And the idea that they're all going to be on one standard, it just doesn't ring true to me. I think given they all have their own systems, standardization is going to be very hard to come by, as much as it is kind of the holy grail but kind of like the target the end. The end goal, I think, given the constraints of the moment, and the constraints of the moment rely on the kind of lowest common denominator. And that common denominator is often just paperwork, right? It's just communication, whereas the method of communication between all these different players, and it's the only kind of standard they have, and we think that's gonna, that's gonna persist. And really, I mean, I could go on for a long time about this, but we're a company we solve problems. And the problem, real problems today, I'm basically saying like, okay, we accept that this is the status quo. How do we help companies do a bit better? And that's, that's really a core belief. In many ways, you can think of us as being like, the universal adapter. We're agnostic, we don't care what comes in, we deal with it. And we help enterprises around the world have taken that nonstructured kind of data and deal with it and analyze it to that format. And I think that's really powerful. I think another great analogy that like I again, I don't want to like belabor this point. But a great analogy is that people talk about blockchain and stuff in the space. And the big problem that that that kind of approach has is what I call the Oracle problem is how do you get something from like, the physical world onto the digital into digital? And I think that's, that's a large part of what we do. We take the status quo, and we bring it into that kind of digital future. And I think that concept is super powerful. And it's obviously something that resonates with a lot of our customers.Santosh Sankar 8:32 Yeah, and putting a bow on the paper bit. When people ask me and try to push on it. The beauty of paper across languages, borders cultures, is that it's the most flexible medium. So as long as you know, we have this fragmentation just around the people’s side of things. Paper is going to be a big part of it. And I love that you know how you talked about being the adapter? And I think something that people often mistake this for is Hey, is it just, you know, OCR? I think it's lost on people what it means to be the adapter. Could you expand on that?James Coombes 9:17 Yeah, yeah, for sure. And I also think, like some of the previous kind of previous points that you've done, you've made right in terms of kind of a universal adapter approach. Or, let me kind of take a step back. What we see like what kind of paperwork aspect is really is really in many ways. It's like the content of a contract that people are kind of passing in between each other. And you see this in so many spaces. You see this in a kind of a swift kind of world and then kind of you see it in GDI. You see that even though you've got a bunch of standards, there's always a bunch of like freeform areas where you kind of need to capture the nuance of that communication on that paperwork. I think that's also something that shouldn't be kind of ignored and forgotten that the huge amount of nuance is involved in the paperwork as well. But we capture. Now to your point on the OCR side. Absolutely. I mean, though, OCR on itself is... it comes up a lot. We talked about it in terms of, it's probably like 5% of what we do with a product. But it is really, it's really important to control. We spent a lot of our time. Actually, I mean, a huge amount of time actually building and training our own AI, so we don't use anything else. We use our own. We train them around on literally millions of images. And we've done it specifically in the supply chain space because supply chain paperwork is always really painful. There are watermarks everywhere. People like to print on top of stuff. We do handwriting so it's important to get that right. And so we trained and built our own so we have control and then that feeds the rest of what we do. The balance and this is a kind of a hard bit is more kind of the extraction site is, is to be able to say, you know what it's not just about extracting the text it's about specifically extracting kind of poor learning is why over contain a number of acts. And doing that in a meaningful way but doesn't involve you know, kind of historical approach are like templates and stuff has to do it in a smart way that can adapt and learn with an operator. And that's kind of where the machine learning part of what we do kind of reality comes into the picture. So that is the lion's share of what we do in terms of the extraction site. But you're right, it's leaps and bounds above OCR, and kind of embedding all of that