Phil Pierce · AI systems architect
I build AI systems that decide correctly.
Twenty years leading the teams and building the systems that make an organization's riskiest calls, inside two of the most heavily regulated companies in America. Inventor on pending patents in AI belief-memory.
Eleven facts. Each one true at some point on the horizontal axis, each one learned at some point on the vertical. The verdict is not a property of the evidence. It is a property of when you asked.
Tools
The spec →Four minutes
Architecture check
Fifteen questions about one of your AI systems. Tells you which of the three properties it's actually missing.
Run it →Bring a CSV
Calibration check
Predictions and outcomes in, reliability curve and calibration error out. Checks whether a system is honest about what it doesn't know.
Open it →Writing
All essays →August 2026
6 min
What Did You Know, and When Did You Know It?
Two questions decide whether a system's reasoning can survive scrutiny. Most AI systems are architected in a way that makes both unanswerable, not by accident, but by default.
Read →August 2026
5 min
A Confidence Score Is Not a Reason
0.87 is a summary of an argument. The summary gets stored. The argument, the part that actually explains the call, usually doesn't.
Read →August 2026
6 min
You Can Buy Detection. You Cannot Buy Judgment.
The judgment layer can't be purchased, because the one input it depends on, what a finding actually means inside your organization, never leaves the building.
Read →The list
New essays, occasionally.
Evidence, memory, and machine decisions that have to hold up. No schedule, no filler.
Contact
Building AI systems that make real decisions? Let's talk.
This is the problem I spend my time on. If you are designing or building one of these systems, write to me.
phil@philpierce.com