hi, I’m Jimmy

I’ve spent years building the systems businesses run on.

Most of it was web development for marketing and ecommerce, which means I spent a long time in the unglamorous parts. Forms, CRMs, order flows, and the integrations between them.

For the last few years I’ve built AI workflows inside a company. Long, tedious processes that ate people’s days. The kind of work nobody defends but everyone keeps doing.

Here’s what that taught me that a demo can’t. I had to live with everything I shipped. When something broke, I fixed it. When people didn’t use it, I found out why.

That’s why I start with one workflow instead of a strategy deck. The projects that survive are small, boring, and chosen carefully.

Jimmy Roman
the unglamorous parts
The kind of work I go looking for

It looked like judgment. It wasn’t.

A selective program receives over a thousand applications a year. Each one read by hand and scored against a rubric across three documents: a CV, a letter of recommendation, and the application itself. Forty reviewers. One to two hours per application. A few weeks of calendar time, every year.

It had never been automated because it looked like judgment. It wasn’t. It was a rubric, applied consistently, at volume. That’s a system’s job.

I built it. Same rubric, same criteria, applied the same way to every applicant. A person still makes the final decision. They just stopped doing the reading a machine can do.

Conservatively, that recovered about $23,000 a year in salaried review time. The build happened once. The recovery repeats every year.

Your business probably doesn’t grade applications. But if anyone there reads a stack of documents and scores them the same way every time, you have that workflow wearing different clothes.

Resume screeningRFP qualificationIntake triageClaims reviewVendor evaluation
1,000+
Applications a year, read by hand
40
Reviewers · 1–2 hours each
$23,000
A year in review time, recovered
Why I do this on the side

On the side, on purpose.

current, not theoretical

I do this work every day inside a company. I take on a small number of outside clients.

That’s deliberate. It keeps me current instead of theoretical, and it means I only take work I can actually finish.

What I believe

Three things, held firmly.

01

Most businesses don’t need more AI ideas. They need better workflow decisions.

02

The best first project is usually boring. Repetitive, high-friction, and done by someone expensive.

03

Technology should make a business calmer, not busier.

the whole point, really
where to start

$999. Forty-five minutes. A ranked plan you can act on with or without me.

Book the AI Workflow Audit — $999

Not ready? Read Worth Automating — a weekly note on which workflows are worth fixing and which aren’t.