AI for
customers
Useful, trustworthy AI capabilities embedded in customer-facing products: retrieval, personalization, conversational interfaces, and the evals that keep them honest.
↗Product and Engineering Executive
Software, AI, and things I’m building.
I’m a product and engineering executive. This is my personal space for writing, experiments, and projects I find interesting.
Latest dispatch article
OpenAI’s Dot Found Me a NYC Apartment. I Just Showed Up.
I gave dot a goal, made decisions along the way, and showed up. It helped with the apartment search, viewings, paperwork, and the little things after signing.
02 / Operating model
Questions I keep coming back to
AI for
AI for
employees
Governed access to models, tools, knowledge, and workflows: model gateways, usage and cost visibility, and adoption without shadow AI.
↗AI as the
engineering factory
Agentic systems that plan, build, test, review, and improve software delivery. The agent hands over a working environment with evidence, not a plausible diff.
↗- 01 repo open-source / self-hosting OpenGym: Own Your Workout Data A self-hosted gym and body-weight tracker with routines, workout logs, progress charts, and data imports. Optional AI coaching sits on top of a useful product.
- 02 repo agents / ai-for-employees OpenWork: Make Agent Capabilities Easier to Share Different AI’s OpenCode-based workspace combines a desktop app with shared skills and MCP connections. The useful test is whether a working setup transfers cleanly to the next person.
- 03 repo agents / verification Playwright CLI: Give Coding Agents Browser Evidence Microsoft’s browser automation CLI lets agents exercise user flows and capture screenshots, traces, console messages, and network requests. Use the evidence alongside regression tests.
- 04 repo evals / agents Promptfoo: Make AI Quality a Repeatable Check A CLI and library for evaluating and red-teaming AI applications, with declarative test cases and CI integration. Start with failures that matter to your product.
- 05 repo agents / mcp REA: Give Coding Agents a Way to Inspect Existing Software An MIT-licensed CLI and MCP server connects coding agents to reverse-engineering tools. The useful question is how well an agent can explain existing behavior with evidence.