Extend

For the past 12 months, we've been in the lab working on Extend and haven't shared too much publicly. That's changing starting today! We’re excited to introduce Extend, an AI platform that helps companies turn their unstructured documents into a competitive advantage. Modern, ambitious companies like Brex, Opendoor, Checkr, Vendr, and more use Extend to turn their messy PDFs, images, and files into new products, happier customers, and faster growth. I’ve long had a personal connection to this problem, experiencing it first hand as an early engineer at Brex in 2018 just as we were launching the initial corporate credit card product. Our goal was to build a magical experience for employees, by parsing receipts and automatically matching them to the correct expense. Sounds straightforward, right? Turns out, since users were uploading these receipts in real time, there were an infinite number of edge cases to consider. Millions of vendor formats aside, the receipts could be upside down, blurry, crumpled up, covered in coffee stains, have handwritten tip amounts, and so much more. Building that was one of our most most complex engineering projects spanning months of implementation, iteration, and maintenance. Not every company can afford to do that, or even should. Ambitious teams want to focus on innovating, not data plumbing and fire-fighting edge cases. And businesses shouldn’t have to choose between speed to market vs. building something in-house that becomes a competitive advantage. Legacy OCR, IDP, and point solutions have existed in the market for a long time. If you can get away with using one of them, you should — we’re probably not the right choice for you. But often times those legacy solutions fall short because: 1. They lack the capability to handle your complex data and requirements 2. They cannot be customized to power the unique experiences that teams envision Our customers are ambitious companies trying to innovate, leverage the latest advancements in AI, and don't want to be tied down by off-the-shelf options. So they decide to build it in-house — and while impressive demos are easy to build with LLMs, there’s a world of difference between that demo and confidently deploying an enterprise-ready, production use case. Extend closes that gap by providing a platform that accelerates in-house builds to production-level confidence in days, not months. Having now deployed robust AI workflows into production on mission-critical use cases at startups and Fortune 500 companies alike, one lesson has become very clear: OCR is dead. The problem is no longer “can we extract text from a PDF?” That’s table-stakes. Rather, the problem becomes: "how can we effectively teach AI models with PhD-level intelligence the intricacies of our documents, business, and workflows in a way where they can drive business impact?". That’s what Extend solves, and we’re the only ones who do it. Full link in comments below!

My identity was stolen and this is what it taught me about B2B sales. 💀 No really, I did actually learn a lot about my product’s benefits through the experience. A month ago, some malicious wannabe HGTV star opened an Ashley Home Furniture credit card using my identity. Not only is that the lamest credit account to have in my name ever, but it also made it especially difficult to dispute. Courtesy of TD Bank’s (the card issuer) 2004 time capsule of a website and 5.5 hours of customer service holding time, I finally got the card disputed. I’ll know – by snail mail – “within 30 days” if my dispute holds water. The same timeframe from all 3 credit bureaus, and the FTC. After such a hellish experience, this whole operation is mission-critical on my priority list to get resolved ASAP – and now I’m in the dark for potentially a month to even hear an update. So what did this teach me about what I’m selling? – I’m not just complaining here. It taught me to see the direct customer-side effects of the problem that Extend, the YC backed, AI platform for processing unstructured data, is solving. One of the major reasons all of this takes so long can be directly tied to the variety and complexity of documents involved in this process. For each dispute (and mind you these are all a particular subset of a fraud case, there are so many different routes a dispute could be in regards to), I was encouraged to upload supporting documentation. FTC claims, police reports, maybe some receipts? PDF’s, images, physical mail. All being thrown at the problem. With no guidelines, let alone standardization, I, and anyone else affected, can toss in whatever we feel like that we believe will help the cause. There are no rules. No standards. Just a digital free-for-all. Until now, there were 2 ways this could be “handled” 1. On the other end of the screen, some poor soul has to sift through this mess, trying to make sense of it all. Manually classifying, filtering, and extracting data from the documents. To be faster, you throw more people at the problem. 2. Traditional automation tools? Useless. OCR? ML? Not a chance. They can handle a few of the most popular formats, but any new document they weren’t “trained” on and you’ve got yourself a need for human review. Now there’s a third way: 3. Built on the latest LLM’s, Extend enables teams to use NLP prompts to classify, extract, and validate any type of unstructured data, regardless of variety or format. Customers like Brex, Opendoor, Vendr, and others are driving faster customer processing times, and higher NPS, by automatically processing documents with Extend, even seeing a 75% reduction in manual tasks. P.S. Anyone got tips on how to avoid getting your identity jacked? Seriously, hit me up. This sucks. #documentautomation #B2Bsales #operationsmanagement #AI #techinnovation #AIinbusiness #businessSolutions