About
I'm an operator who writes.
Fourteen years in regulated financial services — operational risk and collateral controls, then client advisory — taught me how real work holds together. Now I build AI systems for regulated businesses, and write here to think in public about what that actually requires.
Where I
came from
I came up through the back rooms of finance, not the front. Out of the University of Illinois at Chicago I started at Bank of America Merrill Lynch — governing collateral under Dodd-Frank and ISDA, coordinating cross-asset operational-risk reporting, working in the parts of a bank where a number has to be reproducible and a named person has to answer for it.
The surface kept changing; the job didn't. Loan rehabilitation under federal rules, where compliance and a hard human conversation had to hold at the same time. Then more than a decade at Morgan Stanley, where I built a client advisory practice from a standing start to $115M — walking executives and households through the equity-compensation events where a small mistake becomes a large, personal one. Different rooms, same discipline: be the one who is accountable when it has to be right.
That work shaped how I see almost everything. I learned to read an organization by looking at what it protects, to design for the bad day rather than the demo, and to treat ownership as the only control that reliably holds. None of it was glamorous. All of it was real.
When capable AI arrived, a lot of people expected that kind of operational knowledge to lose its value. I've found the opposite. The tools got faster; the judgment about where to point them got scarcer. Knowing which part of a workflow you can afford to be wrong about — and which you can't — turned out to be exactly the thing these systems can't supply on their own.
What I
work on
Through Rios Applied AI, I help regulated businesses put language models to work inside processes that already have rules, auditors, and consequences. The interesting problem is almost never the model. It's everything around it: the controls, the ownership, the failure cases, the people quietly holding the current system together.
Why I
write
Writing is how I find out whether I actually understand something. If I can't make an idea precise on the page, I don't really have it yet. The Field Notes are that process, made public — partly to keep myself honest, partly because the people I most want to work with tend to think the same way.
What I'm
trying to
understand
Right now: where the real boundary sits between what a model should decide and what a person must remain answerable for — and how that boundary moves as the tools improve. It's the question underneath most of what I publish, and I don't think it has a fixed answer. That's why it's worth writing about.