Bots and Beyond Episode 29: Reinventing Document Processing with AI

In this episode of Bots & Beyond, host Wayne Butterfield talks to Rob Delaney who is both CEO at Infinia ML and Director at Carrick Capital Partners.

Every organization has vast amounts of data locked in documents of all kinds, from emails, to PDFs, to presentations and invoices. Some estimates suggest that up to 80% of the data produced by organizations is unstructured, which makes it extremely complex to analyze and derive benefit from. In recent years, machine learning and AI technologies have made huge advances in turning unstructured data into actionable insights that business leaders can harness to make better decisions. Infinia ML is at the forefront of developing machine learning and AI-powered solutions that transform the way enterprises process their data. Listen in to find out more about intelligent document processing and why it matters in today’s data-driven world. 

 

Transcript

Wayne Butterfield

Hi, and welcome back to Bots and Beyond. This is an episode of firsts. Firstly, it’s now 2022 so Happy New Year to everyone. And also this is the first episode I'm going to be recording from my new home in the US. And so of course, it would make absolute sense to bring on one of my American guests, new friend and partner to the firm, Rob Delaney. Rob's the CEO of Infinia ML, a tool that I am uber excited about, having recently been exposed to them during my time here. So, it's a great pleasure to welcome Rob to the show.

 

Rob Delaney

Great to be here.

 

Wayne Butterfield

I know we've got a great set of questions and discussion that we are going to run through over the next 20 odd minutes. You've got a really interesting background that we'll get into. And of course, we'll talk a little bit about Infinia and the role that you're playing in helping clients. But first of all, let's learn a little bit more about you. As is the usual intro for our guests, if you can talk to us a little bit about who you are, what you do, and then most importantly Rob, why do you do it? What's your passion? Why are you doing what you are doing?

 

Rob Delaney

I'm the CEO of Infinia ML. I am also a Director with Carrick Capital Partners, which is a growth equity firm that focuses on software and software enabled businesses and is the majority owner of Infinia ML. In terms of my background, most of my background is on the investing side. So, the bulk of my career has been around evaluating companies, technologies, products management teams, from the perspective of an investor or a buyer. And then a few years ago I recently stepped into the role as the CEO of Infinia ML, which is obviously the other side of the table, so to speak. So as an operator. In terms of what motivates me or why I do what I do; certainly there's a few base emotions right there that are integral I think, to why I do what I do. It starts with a competitiveness. I think everybody who is in a senior role at an early-stage company, usually is a pretty competitive, ambitious person that wants to build something great. That wants to be a part of changing an industry, changing a market. Then the other piece is more on the people side. I get a ton of intrinsic or personal value from working with really smart people, learning from them.  There's really no better place to do that than, in my view, I'm biased, Infinia ML. The team we've assembled, I learn more from them every day than they could probably imagine. That might sound a little backwards as the CEO, but I really believe that the value we get out of our careers is very much based on the people we work with and what we can learn from them, how they can push us and help us to grow. Those are some things that motivate me at a high level. At a more tactical level, really, it's the technology that we work on. So, machine learning, AI. There's a lot of discussion out there in the world, about the technology. And it's kind of a catch, because there's very broad application of machine learning and AI across many, many use cases and verticals. But, for me, it's an honor, it’s a thrill to be working at a company that is on the cutting edge, developing machine learning solutions and delivering those for customers and really changing the way they do business.

 

Wayne Butterfield

It really is an interesting mix, the whole two hats. I'd love to dig in a little bit more about the CEO role versus the director at an investment firm. A: Where do you get the time? And B which do you see as the future? Or do you constantly see this dual role moving forward?

 

Rob Delaney

Starting with A: Where do I get the time? I get that question a lot. I think you adjust to what's in front of you. It may be that if someone asked you, “could you take on both of these roles at the same time?” The answer would be if you really thought about it and tried to plan it out, no. But the reality is, you just take what's in front of you one day at a time. And in doing that it's perfectly manageable. I've actually taken the dual role and really tried to turn it from a potential challenge into an opportunity. What I mean by that is there’s this Venn Diagram of the investing role and the operating role, and there's a big, big overlap between the two roles. And I try to take that overlap and use it as an opportunity, but a differentiator for our company. If you think about the role of a CEO, it's a capital allocator, but it's also someone who's supposed to set the vision, the strategy. My role at Carrick actually enables me to get access to a pulse on the market and what other companies are doing, and what's working; even in unrelated industries, unrelated software companies, what's working, what's not; that I probably wouldn't otherwise have access to as just an operator. There are all sorts of other avenues where that Venn diagram becomes really beneficial, where the overlap in that Venn diagram becomes very beneficial for Infinia, in terms of partnership opportunities, customer opportunities, the resources and connections that are available through Carrick. I actually view it as a big positive. Certainly, some weeks are longer than others, but there's no secret sauce you just you deal with what's in front of you, and you take the opportunity and try to turn it into a big positive for the company.

 

Wayne Butterfield

I must admit, Rob, I’m really disappointed. I was hoping that he'd cracked AGI and that the reason you were able to be Superman in these two challenging roles is that you had already invented artificial general intelligence that did half your half your work for you. So I'm a little disappointed.

 

Rob Delaney

If I had that I wouldn't tell you!

 

Wayne Butterfield

That's a very, very fair point. So, you heard it here first, guys. I totally, totally get the crossover. And as you were talking, I was thinking; had you said I was an investment banker in a traditional investment bank where I was looking at, currency fluctuations and things I'd be like, that’s not really a good crossover. But actually, looking at a market, scanning a market, having access to opportunities to look at new and interesting technology, which then enables you to see the types of challenges that those solutions are solving for, does actually give you quite the advantage over most other CEOs. So absolutely as you were talking, it really did resonate and makes sense. So you must have seen a load of good technologies over the years. Why Infinia? Why have you taken this role at Infinia? What was it about the company, the solution, the technology, the people that made you take this type of role within the business?

 

Rob Delaney

I think if you look at when I did step into the role, so you're looking back a few years, and that was in the relatively early stages of the company. Infinia was, for all intents and purposes, founded in 2018. I think it was founded late, late 2017. At the time, and I think this is the journey we've been on, the company was very much a collection of what I would call really good raw materials. And that starts with the people. I have never seen in my career a collection of talent this strong at an early-stage company. I think when I joined, it was probably somewhere in the neighbourhood of 20 / 25 employees. We're closer to 40 today, but those 25 employees, you would find 3 to 4 people that in terms of the skill set and the expertise matched up with our 25. And so, for me, that's where it all starts. Everything we create is based on the people behind it. And everything about building a business is finding people who fit in on a team and who can execute against a really important function. That was the biggest driver for me, just looking at the team and how can I help steer this team towards a great market opportunity, where they can leverage all their skills in a coherent way. That, plus the underlying technology. At the time Infinia was more of a custom machine learning services business. But if you looked at the library of technology that they had built out and you looked at the use cases that they had delivered for certain customers, you really could see exponential opportunity within the business. So those are the those are the two really, really exciting components for me.

 

Wayne Butterfield

So, we've heard of the name. We know it’s involving ML. We know that there’s quite a few different components as you've described. Talk to us a little bit more about Infinia: What bucket would you put the capability in? And what type of problems are you solving?

 

Rob Delaney

I think if you're going to go with a bucket that most people will be able to identify with, we fit in the intelligent document processing bucket. And I think there is a lot of what we do that does put us squarely in that bucket in terms of the outcomes. But I view us more as something that I would call a data liquidity company. A lot of times we talk about we're applying machine intelligence to knowledge work. And, and for me, if you think about intelligent document processing, and what is the opportunity, most businesses are still organizing themselves largely via a set of documents that have contracts, and org charts, and agreements, and policies. And there's so much latent value that's in those documents, but it's highly illiquid. This is value that at scale, you can't pull out, you can't drive insights on. You can't aggregate unless you're able to efficiently and effectively get that information out of those documents. And documents being a catch all word, there probably isn't a word in the English language that sufficiently describes it. Documents can mean a lot of things. It can mean an email, it can mean a contract or an invoice, the more standard documents. So, I view us as a company that takes all this illiquid data value and makes it liquid, and then enables our customers to do more transformational things with that liquid data. And for those who may not be as financially inclined as others, the analogy being, if you have a bunch of illiquid investments, it's tied up in investments that you can't readily sell, or trade, or get value from. They're more long term and there's a bunch of work that would have to be done to move them. And that's how we view data that is sitting in organizations’ documents. So, our company, our mission, is to take all that illiquid data, latent value, and help organizations drive value from that data and make it liquid. That comes really in two major forms. The first is can you drive some type of automation of a process that involves a document? And the second is can you deliver some deeper insight or prediction based on the data that existed in the documents from that process?

 

Wayne Butterfield

One of the things that really resonated with me thinking about what differentiates Infinia ML to traditional OCR was; the OCR, the digitizing of paper isn't really the end goal. The end goal for all organizations should not be about turning this paper into a digital format. It should be about the data that’s in the newly digitized format. That's the bit we're all interested in, it could be the best OCR tool in the world with 100% accuracy on every single character, but if that's all you do, you're limited. You are limiting your capability; you are not truly delivering on client problems. Because the client problem isn't necessarily about 100% digitization of the document. It's about what is written, what is said, what does it mean. How frequently does that happen? It's about all the stuff that you can get out of the document. And that's what really resonated with me as somebody who has spent a while, solving for client challenges with OCR / intelligent document processing. That's what really got me, it’s that actually the end goal for you wasn't about a perfect OCR. It really was about the downstream, and it's the downstream bit that is what really helps clients and makes business cases. Not just the quality of your OCR capability. Why then, I'm going to be quite provocative here, why is your approach right in this field? I'm hoping I wasn't too leading, by the way, because that wasn't that wasn't my intention. I just feel very passionately about this area as a whole. So, talk to us about why you think the Infinia ML approach is different, is better than maybe what traditional OCR has delivered so far.

 

Rob Delaney

To take a step back, the comment you were making earlier was really insightful around it's not just about “can you digitize the document?” And I think there's a broader comment around AI and machine learning. People talk about AI and machine learning like they’re a product. But the reality is, they're not; they’re tools. So, they're tools to solve an end business problem. So that's the first principle we think about when you think about document processing. Document processing, using OCR plus; and there are other companies out there that are like us in that they'll use OCR plus. OCR plus meaning use OCR, plus use layout and spatial analysis, plus natural language processing, and several other tools to be able to effectively extract, understand, classify data that exists in documents. That's the right way, we believe, to think about this problem; can you handle the complexity beyond just a standardized form? Can you handle the complexity that exists in documents that organizations are using to communicate with each other or to do some type of agreement? And so that's the core utility that we have set out to build, which is a document processing solution that is able to handle that complexity, that has a very diverse toolbox that can deal with different situations or more challenging problems. But ultimately, as we talked about earlier, that's an automation utility. That gets you to a point, and this point we call it data liquidity, and then our approach is to enable us and our customers to deliver even deeper insights on that data that has now been made liquid. But you could never do that if your approach was just OCR. Because if your approach is just OCR, or the more basic document processing methodologies, you're not going to get the complexity out of the document that you need, to drive the insights. You're not going to get insight. You're not going to get, spatial or classification. You're just going to get, the text that we're able to pull off, we can digitize that and then read it. So that's fundamentally why we do things the way we do. There's another belief that we have internally, that it doesn't really make sense for companies at this point to spend time on making OCR better. So, we view OCR as somewhat of a commodity. And it's really what can you do beyond that? What can you do when OCR isn't effective? What can you do in addition to OCR that then sets you up to be a company that can drive more insights or more transformation? And if you think about our company history, how did we get there? It wasn't just a philosophical approach; you have to look at what is your skill set. And our skill set at its core, our founders, were folks who all were doing research in a Duke Machine Learning Lab. So, at its core, we're a machine learning company that is able to build and deploy more complicated complex models that that can tell you something more than just what you can figure out from OCR. So, long winded answer, but that's, that's how we come about it philosophically and also tactically.

 

Wayne Butterfield

I think that really helps between the traditional OCR and your approach, concentrating on the solving for business problems and client challenges. Where next for you guys then? Because OCR and IDP; I pigeonholed you there earlier on, or asked you to pigeonhole yourself there earlier on, but really ML is far broader than just solving for IDP challenges. So, where next for you guys? And again, you don't have to give away any trade secrets, of course; but are there other areas where you think well actually, that's complementary, or actually that's making use of similar models to what we already have, or actually, this is a natural progression after IDP? Do you kind of have any insight in that? Where next for you guys?

 

Rob Delaney

There are a couple things we think about. So first is IDP and what we do from an automation perspective, fits into a broader basket of full-scale automation of processes. So there are a number of things that probably need to be considered when you're trying to increase the amount of automation in a workflow, in a process, in a business. That includes things like partnerships with, more consulting design services partners, partnerships with RPA vendors. And so, as we continue to advance our technology, we also continue to try to build our alliances with folks who are complementary to us, and who can deliver an end solution for the client, or the customer, that is leading to a higher scale of automation than just you can solve this one little piece of my workflow. So, continuing to build out those alliances and continuing to advance our general document processing capability will allow us to do that. One of the one of the things that is a strength for us today is the modularity of our platform, which allows us to deal with more varied problem types and leverage our toolbox. And then the other strength is this insights piece, our ability to deploy business specific, customer specific insights, or predictions on top of the platform. As we think about the future, the company, that is going to evolve into us getting pretty vertical. And what I mean by that is what we call towers, because it may not be fit cleanly into what most people would describe as a vertical, but we call towers of off-the-shelf insight, predictive models. For us, that's the future as we continue to advance this core automation capability, and then what can we build on top of that, that is highly repeatable across a broad customer base.

 

Wayne Butterfield

Rob, great to hear your views and I really get the feeling that this for you is very much the start of the journey, rather than the end. And having been in America now for about 10 weeks, I don't think I've ever signed as many pieces of paper or had clients talking about the number of faxes and things that they are dealing with. So IDP absolutely is needed, here I think even more so than what I had seen in Europe. But again, it's only the start and really, it's the information within those documents that's really interesting. So, you guys being absolutely focused, and laser focused on making the most of what's in the docs versus digitizing the docs, is very commendable. And of course, incredibly, incredibly useful. And for me, the approach is pretty unique in the in the market.

So thank you ever so much for your time, I know you're an incredibly busy man. So thank you very much for joining us today on Bots and Beyond.

 

Rob Delaney

It was my pleasure, Wayne. I appreciate you having me on the show and as always, I enjoy our conversation.

 

Wayne Butterfield

Thanks for listening to Bots and Beyond. If you liked this episode don’t forget to like, comment and share with your friends and co-workers. We’ll be back in a couple of weeks with a great guest, fantastic and of course real-world insights into the world of automation transformation.