Writing
Field notes on multi-agent systems, choosing models that fit, and getting AI into production. Some pieces first ran elsewhere; those link back to the original.
2026
Building a practice customer: persona agents, hidden objectives, and a judge you can trust
How Customer Scenario Simulator II works: one agent plays a customer who is holding something back, a second grades the conversation against a rubric, and a small eval harness checks that the grader deserves to be believed.
Deep diveYou don't always need a frontier model. You need one that fits.
The short version of my Medium piece on right-sized, open-weight models: the real trade is generality against fit, and the only way to settle it is an evaluation on your own workload.
Companion
2025
Notes from building the agent behind "Four steps for startups to build multi-agent systems"
My colleague Oluwamayowa Awojuyigbe wrote up a multi-agent build I helped architect. These are the four lessons from that project I now repeat to almost every startup I work with.
Companion
2024
Vertex AI Studio or Google AI Studio? Pick by stage, not by feature list
A companion to my most-read piece for Google Cloud for Startups. The products have changed since 2024, but the way to choose between a prototyping sandbox and a production platform has not.
Companion