Applied AI · Product Builder · Mindful Technologist
I help teams decide what to hand to AI, and how to know it worked.
Five years of AI strategy, applied AI and training inside real organizations.
50+
Organizations
20+
Industries
65+
AI trainings and talks
Selected Work
What the problem was, what I decided, and what happened. Including the parts that didn't go my way. Organizations stay anonymous.
- 01
Internal tool at Cadre AI
An Eval for Coding Agents, and the Three Bugs I Found in It
A harness that grades a model's fix against hidden tests in a sealed container and checks whether it passes on every try, not just one. The most useful result was the three bugs I found in it.
- 02
Investment office of a large institutional investor
Choosing an AI Platform for an Investment Office
I turned 18 loose use cases into 20 workflows a vendor could be tested on, held six vendors to the same seven gates, and laid out two paths. The executives chose the one our deck didn't lead with, and two follow-on contracts were signed.
- 03
My personal context layer, and Canon's design
Knowledge an Agent Can Act On
I rebuilt my personal context layer so every claim is tied to the exact words it came from and checked by code before an agent can use it. Then I designed Canon, the version for a whole company.
- 04
Internal product, used with the firms we trained
A Personalized Claude Coach and the Meta-Skill That Builds It
A Claude coach built for each firm we trained, and a meta-skill that builds one for the next firm. Used in live training at eight firms, and set up end to end by a colleague with no help from me.
Latest Writing
The Great Inversion
Your org chart was designed for expensive execution. That execution is getting cheaper every month, and the org chart has to change with it.
8 min
The Factory Floor
Follow one order down a production line staffed by AI agents, then turn the dial and watch where the people go. A scroll-through explainer of the software factory.
12 min
Robot in a Room
One robot, one sealed room, fourteen steps. A scroll-through explainer of how a chatbot becomes an agent.
10 min
How I Work
People First
My first startup taught seniors to use technology. One of them, Norma, learned to video-call her grandkids for the first time. I still judge AI work the same way: does someone's day get better, and can they keep using it once I'm gone?
Define "Worked" First
I write down the result I expect before every run. In a financial-aid tool, every dollar figure came from fixed math, never the model. If a team can't say what good looks like, it isn't ready to hand the work over yet.
Fix the Boring Thing
At a federal supplier, a quarter of order lines never got a vendor response. The causes were an unwatched inbox, an email that looked like spam and an attachment that opened as "file corrupt." None of it needed AI. Discovery earns its keep when it finds that.
Show the Trade-Off
At an investment office I gave the executives two honest paths. They picked the one our deck didn't lead with. Because the trade-offs were already on the table, nothing had to be redone.
What People Say
That was the best presentation I've ever seen about our workflows. No one from outside of our company has put it together like that.
Coming in pretty much cold, you got us from nothing to pretty capable. Now we're off and running and building things.
You in particular, Ben, you've done a great job of listening and building an understanding in a short period of time about a lot. And that was evident.
This is far beyond what we initially discussed. I had really limited, tiny use cases. This is way cooler.
Very few teams get to this level of care and ask these questions. You're asking the right questions.
I grew up on stage, in plays and musicals at an arts high school. It's still the most useful thing I bring into a room: reading an audience and telling a story that lands. Then startups, private equity, and co-founding an AI consulting firm. None of it came from business school.
Learn more about me →
