Measured
Output is scored against a standard you set, so you have a quality number — not a vibe. The moment something is measured, drift can’t hide.
About • Curio Chat Academy
CurioChat is what happened when someone from that world looked at “AI for business” — and saw toys.
02The Standard
For 35 years, I built the systems banks can’t afford to get wrong.
Market-risk engines. Derivatives pricing. The regulatory-reporting infrastructure sitting behind tens of billions in treasury assets, across two of North America’s largest banks. Real-time trade reporting under Dodd-Frank and EMIR. Liquidity reporting regulators hold banks to.
If a risk number was off, or a report was late, the cost was measured in millions — and in regulators.
That work has one defining standard: measured, auditable, must-not-fail. Not “works on my machine.” Not “looked great in the demo.” Works for everyone, every time, even when you’re asleep.
Thirty-five years of that standard becomes how you think. You stop being able to un-see the gap between it works and it’s reliable. That gap is where most projects die. Crossing it is the actual job.
03What I Saw
I retired in 2026 and turned that lens on “AI for business.”
I saw toys.
Every product was the same shape — a static library of someone else’s prompts, with a quarterly refresh. Nothing measured anything. Nothing learned from the person using it. Nothing let the owner actually understand what they owned.
It wasn’t built like a tool. It was built like a demo: impressive for five minutes, drifting by week three.
And here’s the part that bothered me most. Capable, disciplined people were handed these things — and blamed themselves when the output came out inconsistent.
It was never them. It was the build. You can’t out-discipline a tool that has no way to get better.
04What I Built
So I built the thing I wished existed: an AI system run the way I’d run a bank system.
I built it for my own work first — and it compounded. It got sharper the more I used it, because it was designed to keep what I taught it instead of forgetting it overnight.
That system became CurioChat. Four properties separate it from a pile of prompts:
Measured
Output is scored against a standard you set, so you have a quality number — not a vibe. The moment something is measured, drift can’t hide.
Owned
The intelligence lives in your business, version-controlled and readable — not rented from a vendor who keeps the value.
Self-Improving
Every correction you make is captured and promoted into permanent capability. You pay for a lesson once; the system keeps it.
Auditable
You can see what it did and why — so you can trust it with work that matters.
And because you stay in the loop — correcting it, reading it, owning it — the system gets sharper without you getting duller. Smarter tool, sharper owner, from the same work.
One decays. One compounds. CurioChat is the kind that compounds.
05The Mission
CurioChat Academy exists for one idea: help solopreneurs build a business that learns.
Engineering-grade AI you own — built to the discipline of a bank’s risk system, made teachable for one person. Not a prompts course. Not “replace yourself with AI.” My definition of dependency is simple: the tool gets better while you get worse. CurioChat is built for the opposite — leverage that compounds, while you stay sharp.
The rigor is borrowed from banking. The effort is one workflow at a time. You don’t need to write a line of code.
06Who It’s For
For software engineers who want AI that ships production code to a real standard.
For software engineers →
For business builders who want an AI operating system they own, not a pile of prompts to babysit.
For business builders →
For daily AI users who want the leverage without losing the part of the work that’s them.
For daily AI users →
Not sure yet? Start with the blog — it lays out the whole argument, one piece a week.
07Still Building
Most people retire and stop. I kept building.
CurioChat isn’t a theory I wrote down once. I build with AI every day — production software, on the same measured, must-not-fail standard. The system gets sharper every week because I’m still using it, still correcting it, still banking what it learns.
That’s the whole point. A business that learns isn’t a slogan. It’s a design spec — and it’s reachable this week, by someone who has never written a line of code.
There’s no right moment to start. There’s the moment you stop waiting.