By super.AI
super.AI spent Money20/20 USA 2026 (October 18 to 21 at The Venetian in Las Vegas) at our kiosk, giving live demos to people from banks, lenders, credit unions and fintechs. The best conversations were with people who already had a document project in motion. Five themes kept coming up: fraud, where sensitive documents go, credit union lending, buyers at very different stages, and the work that starts after extraction.
super.AI publishes this article and makes document workflow software. The themes below are our own observations from conversations at our kiosk. We have not named the people or companies we spoke with.
Our CEO Ben Kwon, Account Executive Don Wright and Solutions Engineer Maxwell Warner represented super.AI on the show floor. Here is what they heard.
Fraud came up in nearly every financial services conversation we had. Lenders want to know whether a pay stub or W-2 is genuine before anyone relies on the numbers in it. For them, document processing and fraud review are not separate projects: a fast, accurate read of a doctored document is still a bad outcome.
If you are evaluating document automation for lending, ask early how a tool surfaces documents that need a closer look, and where a person steps in before a decision is made.
The second question we heard most was about data handling. A number of people told us they will not send W-2s and pay stubs to the major AI model providers, whatever those providers say about how their models are trained. For borrower documents, data handling is part of the first conversation, not a security review at the end.
The questions buyers are putting to every vendor:
Some of our most useful conversations were with credit union lending teams. Member loan applications come with the same pay stubs and tax forms that banks handle, and the leaders we spoke with described use cases they need to solve now, not on a future roadmap. A few of those conversations went deep enough to work as discovery calls on the spot.
Some people we met were starting from scratch. Others had already tried to build document automation in-house, and it had not worked. Our takeaway: whether you are starting fresh or replacing a build that stalled, the useful first step is the same. Get one document type running reliably, measure it, then expand.
Reading fields off a document is table stakes. One software vendor in the lending stack described its product as picking up after extraction is done. In our view, the value sits in everything around it: checking documents against each other and against other systems, routing exceptions to the right person, and keeping a record of what was decided.
That is the part super.AI Flows, our workflow intelligence platform, is built for: automating the whole document workflow, with validation built in. For more on where extraction alone falls short, see why invoice and PDF data extraction fails.
Thank you to everyone who stopped by our kiosk, and to the Money20/20 team for a great event. We had great conversations with leaders across banking, lending, credit unions and fintech, and we are following up with the people we met. We look forward to next year.
The main AP automation benefits, with sourced benchmarks for cost per invoice, cycle time and exceptions, plus how AP automation works and best practices for a rollout.
A buyer's guide to invoice scanning software: seven invoice OCR tools in alphabetical order, who each one fits, and how they compare on trials, pricing, and deployment.
Intelligent character recognition (ICR) reads handwriting on forms and checks. Learn how it works, how it compares with OCR and AI extraction, its limits, and how to test it.
