AI Greg: a public MCP server that answers for me
One brain, five channels. Anyone's AI agent can interview me; anything it cannot answer goes to the real Greg with a human approval gate on every write.
AI enablement and adoption
Most companies have bought AI. Almost none have changed how anyone actually works. Closing that gap is what I do: build the systems, teach the people who have to use them, and write down what it takes to make AI safe enough to trust.
Build
Teach
Selected work
One brain, five channels. Anyone's AI agent can interview me; anything it cannot answer goes to the real Greg with a human approval gate on every write.
A grounded tutor for corporate training alumni. It answers only from my course materials, in my voice, with citations, so the training keeps working after the room empties.
A pipeline that turns a Zoom class recording into teaching findings. Deterministic facts first, a redaction gate sized to the measured risk, and a byte-identical approval before anything leaves the machine.
Ask
The box below is wired to AI Greg, a system I built: a public MCP server that answers from my writing and hands anything it cannot answer to me. It is not me. It says so. Try it.
aigreg is the AI front door for Greg Dyche. It is not Greg. Anonymous questions get public-depth answers. When it cannot answer, it asks for an email and the question goes to the real Greg, who replies himself. Nothing is sent to him until you confirm the email.
Before this
Data Center Director and Chief Information Security Officer at a health system, with 40 staff and a $19M budget. Client Executive at IBM. VP of sales at a private aviation company, where the team I hired and trained doubled across 14 cities. Air Force linguist before any of it. Every step was about getting people to actually use something.