 Just about a year ago, governments around the world forced shutdowns of their respective economies. We'd never seen anything like it. Central banks took immediate action and affected a monetary policy like none we'd ever seen before. They dropped interest rates to near zero, injected a huge amount of cash into the system and they fueled this liquidity boom to support those individuals and businesses that were in greatest need. Banks were overwhelmed with the volume of paperwork. For instance, small business PPP loans and other things, home buying boomed as mortgage rates hit all-time lows. For several weeks in the spring, it was complete chaos but the tech industry stepped up and accommodated work from home. Cloud infrastructure was spun up instantly as access to data centers was really restricted and SaaS companies became a fundamental staple of not only keeping the lights on but helping customers thrive in the face of a pandemic. Automation became a mandate. As humans, they couldn't possibly keep up with the tidal wave of demand and document overload that was hitting the system. Now one of the companies that was there to help financial firms in particular get through the knot hole was Oculus, a company that focuses on intelligent automation to deploy the power of machines to allow humans to focus on what they do best. Hello, everyone and welcome to this CUBE conversation. My name is Dave Vellante and we're profiling the most interesting SaaS startups that are reimagining how we work. And with me is Sam Bodley, the co-founder and CEO of Oculus. Sam, welcome to theCUBE, first time. Hey Dave, thanks so much for having me. I decided to have the conversation. Yeah, me too. So listen, I know you've told this story a zillion times but I want our community to hear us. How and why did you start the company? For sure. So when I was in college, I was having a conversation with my dad and he was telling me about a meeting he just had with his elder law attorney. And the elder law attorney was complaining about having to review hundreds or thousands of pages of financial documents for every long-term care Medicaid application. When you apply for Medicaid coverage to enter a nursing home, you're required to submit 60 months of financials along with your application. And traditionally, the elder law attorney or a nursing home would review those documents literally page by page, line by line to find high value transactions, transfers and other financial trends. And when I heard about this, it just, it didn't make sense to me. I said, you know, why in this day and age isn't there, why isn't there a technology solution that can ingest the documents and spit out a digital report replacing the cumbersome manual page by page review? So it really just started as a research project trying to learn more about optical character recognition which is the technology of transforming images into text. And, you know, as we kind of kicked around different products in the market, we realized that there was an opportunity to build a unique platform that could ingest documents of any format or quality and produce perfectly accurate results. And that was the, the genesis behind what ultimately became Oculus. You were a young man at this time. How old were you at that time? I was 22 when we started. So fearless. And now my friend, Eddie Mitchell, started a company about 20 years ago. We hacked together a Dell system and this camera. It was all about the modern operating room of the future. And he showed it to a doctor and it was just a prototype. She said, how much? And he said, 10 grand. She wrote a check right there. You have a similar story. I mean, how did you see the company? So we do have a pretty similar experience. Our concept was, we want to give perfect results to the customer every time. So if a customer sends us a clean bank statement from Chase or a blurry cell phone image with someone's thumb in the picture from a community bank in Maine and it's rotated sideways or upside down, like we want to give consistent, perfectly accurate results every single time. And our view was to completely solve the business problem. So the very first version of the software that we built, we had a rudimentary machine process to extract 60 or 70% of the data. And then we had a little tool built on the back end where literally me, myself, and some of our early employees would clean up the data output, make sure it's perfect. And then when we clicked submit, we'd return to the customer accurate data that could be used at the time for a Medicaid decision. And what happened is while we were in our beta period, customers fell in love with the product. They felt it was magical and really just superior from an accuracy standpoint to anything they'd ever tested before. And one of our beta testers said to us, where do I submit credit card information? So at that time, I turned to my colleagues and I said, I think we're ready to, I think we're ready to start charging for this thing and roll it out for prime time. Now, when I was researching the company, I learned that you leveraged, at least some of the idea came from the Captcha. And I never knew this, but the Captcha that we all hate came from Google where they, right, they had, at one point you could, maybe you still can, you could go online, you can read books and have to, it's just scans, you can't even read the stuff half the time. So they were putting the Captcha in front of us, quite brilliant to try to solve for those white spaces that they didn't know. So how did you learn from that? Was there an API that you could plug into Google's dataset or did you do your own? What was that? How did that all work? The concept of human in the loop is super powerful, right? So when we first started, we recognized that OCR and machine data Captcha couldn't do the job completely. OCR, generally speaking, can process financial documents with roughly 80 to 85% accuracy plus or minus. Machines particularly struggled with semi-structured and unstructured documents where the format is unpredictable, as well as lower quality images. So pretty early on, we recognized that we needed human intervention in the process in order to achieve perfect accuracy every single time. And also to create training data to constantly teach our machine learning models to get smarter and drive additional automation. So as I mentioned, the very first version was myself and other employees verifying the data on our own. But as we started thinking about how to scale this up and take on millions and millions of documents, we needed to learn how to better parallelize task and really build the system for efficiency and for scale. So we learned about the Google Books Initiative and their ability to leverage Captcha technology and a distributed workforce to verify pieces of information that their systems couldn't automatically read from books. And we took a lot of those learnings into building our human in the loop infrastructure. And a way to think about our product is it's the marriage of machines and humans that makes us unique. As much of the heavy lifting as we can do with machines we do, but whatever we can't do automatically, we slice into smaller tasks and intelligently route those tasks to humans to perform verification. We then layer in algorithmic checks to make sure our humans did the review correctly. The customer gets perfect results and that same perfect output is used in a feedback loop to train our machine learning models to get smarter and smarter, which dynamically improves the product on an ongoing basis. And the folks at Google were onto this pretty early with the Captcha technology and we were following in their footsteps with our own unique take on it, but specifically applying it to financial documents. Yeah, I mean, on theCUBE, we know a lot about this because we're looking at transcriptions of video all the time and it just keeps getting better and better and better and our systems get smarter and smarter and smarter. So we're sort of closing that gap between what humans can do and machines can't. And I would expect that you're seeing the same thing. I mean, do you think there's always going to be kind of humans in the loop in terms of the quality or is that gap going to be six nines in the near term? I think it's going to take a while to get rid of all the edge cases. You mentioned the PPV program, like we've been on the back end processing PPP loans for some of the leading players, like Cross River Bank, Bluevine, Square Capital and others. And what we've seen during the PPV process is just a wide variety of different documents and inputs and a lot of difficult to read documents that are very challenging to automate. So I think we will incrementally continue to automate more and more of the process, but the value of having humans plus machines is much more powerful than just having machines alone or just having humans alone. And as it relates to the end customer, our goal is to do as much of the mundane work as possible to free our customer up to do the more cerebral analysis. So in a lending context and for the record, our biggest market opportunity is in the lending space. Despite the fact that we started with Medicaid attorneys, we quickly pivoted and realized that our technology was super valuable to lenders to help them automate the underwriting process. And our thesis is if we can take out all of the necessary evils, like document review and allow underwriters to focus on the actual analysis of financial health, it's a win-win-win and creates a really, fantastic complementary relationship between us and our customers. Yes, I want to ask you about the pivot to financial services. You said you started, well, you had the inspiration from elder law, it was Jimmy McGill, AKA Saul Goodman, Breaking Bad, you got started in elder law, but then you made the pivot to financial services really pretty early on. I mean, you had really good, great product market fit, but you kind of went for it. I get an early 20s, you didn't have a big family at the time, didn't have a lot of risks. So you went for it, right? But talk about that pivot because a lot of companies wouldn't do that. They'd get comfortable and just, you know, stay where they're at, but you made that call. It was a big risk for sure. I mean, look, the product was working. We launched the paid version of our product in 2016 and pretty quickly we were onboarding dozens of accountants and attorneys doing Medicaid work. In mid to late 2016, we got introduced to a large small business lender in New York City called Strategic Funding Source. They've since renamed their company Capitus as the current name, but we met with the CEO and the head of product and showed them a demo of the technology and they said, you know, quote, unquote, we've been looking for this for years. We've been looking for something exactly like this for years. And we said back to them about how many pages of financial documents do you review every single month? They pointed out to a bullpen of dozens of people sitting there tearing through bank statements, page by page, line by line. And they said, you know, it's hundreds of thousands. My eyes almost fell into my head. I couldn't believe the volume and it was much bigger than what the, you know, single accountants or attorneys were doing. So we made the strategic decision to pivot at that time and focus on fintech lenders, continue to tailor the product and build additional features for the fintech lending space. And, you know, lending in general had the perfect mix of short sale cycle and high average customer value that allowed a company like ours to scale and ramp our revenue quite quickly. And then the other thing that happened is kind of as we were getting deeper and deeper into the space, the fintech space as a whole started growing massively. So we kind of had the perfect storm of product market fit plus the market growing that allowed us to really ramp significantly, grow revenue. And, you know, despite the fact that it was a risk, it was totally right decision to focus the business on financial services. Yeah, much bigger, Tam. And you could subjectively measure it by the size of the stack of papers. How does this relate to RPA? It's, you know, the RPA hot space. You probably get this question a lot. It sounds like there are some similarities with software bots. What's the similarity? What's the difference? Good question. It's totally a synergistic offering, right? So RPA companies like UiPath and Automation Anywhere, they typically provide a horizontal toolkit to allow, you know, banks and lenders to automate much of the mundane work. Like for example, collecting information from emails or doing onboarding for a new employee or, you know, different types of tasks that a manual worker would have done but can be automated with relatively simple code. What happens in RPA workflows is they get hung up on tasks that can't be completely automated. So for example, a robot might be, you know, trying to complete an N10 lending flow, but when a bank statement is submitted as part of that flow, the robot can't parse it. So what they do instead is they route it to an underwriter who performs a manual analysis, keys information into a back office system that a bank is using, and that information then gets handed back to a robot and continues the automation flow. What we do is we plug the gaps that used to be manual. So a robot can pass us documents like bank statements or pay stubs or tax docs. We run our unique human loop process. We return structured JSON output directly to a robot and it continues into the, you know, to the next step of the flow and, you know, in summary, the combination of robotic process automation and human in the loop, which is what we're doing creates true N10 automated flows rather than RPA might by itself might get you 80% of the way there. So do you have like software integrations or partnerships with those companies? How are you integrating with them? We do, we have software integrations with both UI path and automation anywhere. In our core FinTech lending business, RPA isn't as prevalent, but we are now expanding beyond FinTech lenders into mortgage lending and traditional banks. And we're also, you know, expanding use cases, right? Like historically, small business lending was the core of our business. More recently, we've moved into consumer auto, mortgage, additional asset classes. And as we've gotten deeper with financial institutions, we've seen even more opportunity to partner and coexist with broader RPA players. Yeah, great. I mean, I was just staring at their S1, I guess it was, I think it came up Monday. Over half a billion dollars in ARR, they're actually cash flow positive as UI path. So we're going to see them hit the public market shortly. Hang on, folks. Now, so, okay, so this is, you sell a SaaS, right? A SaaS service, even though there's that human in the loop, that's all part of the service. How do you, how do you price? So you stitch-based model. So we kind of try to model ourselves to serve a massive company is super powerful. We apply that same concept to document processing. So it's a usage-based model. Customers will pay us either per application, per document or per page. And if they want to subscribe for one document per month or millions of documents per month, it's up to them. And we're able to flex up and flex down to meet the supply and demand. And that concept, that scalability and flexibility was particularly powerful in the PPP program, right? PPP was kind of a very unique situation in the sense that lenders weren't able to predict the amount of loans they needed to process. In normal lending, a small business lender can tell you, hey, we expect to get roughly 10,000 applications in the month of April. With PPP, they could tell us, hey, we're gonna send out 200,000 marketing emails and we expect 30% of people might reply, but we really don't have any idea, right? So what happened is the big banks ended up hiring, without exaggeration, thousands of temporary employees to come in and review documents and kind of scramble to do this in a work-from-home setting during the pandemic. Whereas Cross River, they took a technology-first approach. They implemented our API in the back end and it enabled them to instantly scale up their resources. And the result of that is Cross River ended up becoming a top three PPP lender nationally, outperforming many of the big banks with a super-efficient and fast document review process because we were able to help them on the back end with the automation. That's awesome, I love the pricing model. You mentioned Amazon, is that the cloud you use or? We do, our product is hosted in AWS and we take a lot of learnings from them from a business model and positioning point of view. Yeah, and I'm thrilled to hear you say, I mean, I think a lot of forward-thinking startups are doing the consumption model. I mean, you certainly see that with companies like Snowflake and Datadog and Stripe. I mean, I think that that SAS model of, okay, we're going to lock you into a one-year, two-year, three-year term. Sorry, if you get acquired, you're stuck with some stranded licenses, that's your problem. I think that you really thought that out well. You mentioned you're sort of expanding your total available market now, looking at new markets. What are some of the big trends that you want to ride in the coming decade as you scale your company? The biggest one for us is mortgage automation. The kind of V1 of FinTech, small business and consumer loans were optimized and we went from a place where you would deal with a loan officer and have an in-person transaction to modern day, you can get a loan from a small business. If you're a small business, you can get a loan from PayPal online, effectively instantly. If you're a consumer, you can get a loan from SoFi or Lending Club, super smooth digital experience and really revolutionized the way that the market thinks about financial products. I think the next wave of that is mortgage and that's what we're focused on. Mortgage is a massive market in the sense of thousands of lenders, the average application contains a couple hundred pages worth of financial documents and the pain points of the backend of the mortgage process were really accentuated during COVID, right? Refi volume went way up and mortgage lenders were forced to process that volume in a work from home setting. So what happened is mortgage lenders were struggling with the concept of sending personally identifiable financial information to underwriters who aren't working in an office. They're working at home and, you know, kids running around and a million things going on and it's just more difficult to manage than ever before. And, you know, as the volume kind of normalized a bit and mortgage lenders thought about their own future of automation, I think there was just clear recognition across the board that these mortgage lenders needed to learn from some of the fintechs and really focus on automating the back office piece. And, you know, to your point earlier about business model, what we think about is translating costs that used to be a fixed cost and turning them into a variable cost. So now instead of worrying about having to match supply and demand and hire or fire people depending on the volume that's coming in on any given month, a mortgage lender can instantly flex up or flex down and have a super fast, accurate process to handle the docs. And, you know, we're seeing just awesome traction in the market with that, with the mortgage space and we're excited to push forward there. That's great, thank you. I mean, Sam, you described the chaos that worked from home. The financial industry is very overly officious, if you will, it's very security conscious. How do you handle security? Maybe you could comment on that, how you think about that. Sure, I mean, we take a compliance first approach. We built the product from the ground up with compliance in mind, knowing that we were selling into financial institutions. We have a SOC2 type one and type two certification, which is, you know, an industry standard. All of our verification happens with Oculus employees. So there's no third parties involved in our process whatsoever. And then lastly, but perhaps most importantly, our product in and of itself is innately, you know, innately drives compliance. So every data point that we process from a financial document, we not only return the data, we return an exact bounding box coordinates of where that data field appeared on the original source. So that audit trail lives with the loan throughout its life cycle. What we saw prior to Oculus is a mortgage would go through an underwriting process, they'd make a decision, and then that loan might be sold downstream and a diligence firm has to come in and they don't have the resources to review all the loans. So they review 15% of the loan tape and then they say, you know, they give a rating. And what we do is we proactively tackle that by creating a perfect audit trail upon origination that can live with the loan throughout its life cycle. And that process and that traceability has been, you know, super valuable to our mortgage and banking partners. Yeah, so you can ensure the provenance there. So let me end just by talking about the company a little bit. So you incubated, you nailed the product market fit and you pivoted and you re nailed the product market fit. And like a lot of companies in your position, I would imagine you saw your growth come from just having a great product, you know, initially word gets around, but then you got a scale. Maybe you could talk a little bit about how you did that, how you're doing that, you know, where you're hiring, how you're hiring, what your philosophy is on scaling. Sure. Look, I think the key for us is just surrounding ourselves with the right people, you know, the right mentors, advisors and investors to help us really take the business to the next level. You know, we had no pride of authorship when we were building this and recognized that there were a lot of people out there who had been there, done that and can really guide us and show us the way. I know you had interviewed Mark Roberge on the show previously, formerly the CRO of HubSpot. Mark was someone that we, you know, we read his book and had taken sales advice from him from an early age. And over time, we got him a little bit more familiar with the company and ultimately Mark and his partner Jay Poe at Stage Two Capital ended up investing in Acralis and really helping us understand how to build the right go-to-market engine. As the company got bigger, we took on investments from really reputable firms in the financial services space. So our largest investors are Oak HCFT, FinTech Collective and QED investors. You know, QED was founded by Nigel Morris who was the co-founder of Capital One. They backed SoFi and Prosper and a lot of the big FinTech lenders and, you know, bringing the collective expertise from the FinTech sector as well as, you know, from a sales and go-to-market strategy point of view created the right mix of ingredients for us to really ramp up significantly. We had an awesome run over the years. We were pretty recently recognized by Ink Magazine as the number one fastest growing FinTech company. And, you know, as the momentum has increased and the market conditions have been very favorable to us, we just want to double down and expand. Mortgage is the biggest area of opportunity for us. And what we're seeking from a hiring perspective is, you know, go-to-market sales, account executive type resources on the mortgage side, as well as, you know, deeper products expertise, both on the mortgage side, as well as with machine learning, you know, our product because we have the human and the loop piece, we create massive amounts of training data on a daily basis. So it's a, I think a really exciting place for cutting edge machine learning developers to come and innovate. What can you share with our audience about, you know, your company, any metrics and whatever you're comfortable with, how much money you've raised, how many headcount, if you want to get some companies a comfortable given ARR or others on it. What can you share with us? Sure, you know, we've raised about 50 million in venture capital. We have grown from one to north of 20 million in revenue in the last three years. So particularly since, you know, 2017, 2018 is when we really started to see the growth take off. Company size, we have about 800 to 900 employees globally now. We have about 200 corporate employees who perform the, you know, the day-to-day functions of Oculus. And then we have a long tail of about 600 or so verifiers who perform data verification and quality control work. Again, speaking to the human and the loop piece of the bottle, we're, you know, we're focused on expanding beyond the fintech customer base where we serve customers like plaid, PayPal, lending clubs, SoFi, Square, et cetera into the mortgage space and ultimately into the traditional banking space where, you know the problems frankly are extremely similar just on a much larger scale. Sam Bobley, congratulations on all the success. You got a great road ahead and really appreciate you coming on theCUBE. Dave, thanks so much. It's been a great chat and look forward to keeping in touch. All right, did our pleasure. And thank you for watching everybody. This is Dave Vellante for theCUBE. We'll see you next time.