Product. AI. Developer experience.
GARETH
PAUL JONES.
Product Lead, AI at Poe.
I work on turning model behavior into useful products. My background is in APIs,
developer tooling, and getting technology into people's hands.
Watch my talk on evaluating AI agents ↓
THE WORK IS THE TEST.A practical lens on agents: evaluate the work we
actually expect them to do.
FROM THE PRACTICE.
Ideas developed in the open.
Featured talk
Evals for
AI Agents
A practical look at testing agents against the work we actually expect them to do.
Watch the full talk ↗
01
Define the work
you expect the agent to do.
02
Look at the behavior,
not only the demonstration.
03
Use evaluation to ask
more useful questions.
THE LONG
GAME.
Experience across product, research, engineering, and developer relations. Different
teams; a continuing focus on usable technology.
AI product work · 2023–now
Poe / Quora & Zoom
Leading AI product work at Poe since 2024. Previously AI Companion, speech, and
applied ML at Zoom.
Developer experience · 2020–2023
Twilio
Led the API platform, SDK, and CLI developer experience as Group Product Manager.
Product & engineering · 2016–2020
Foursquare
Led the Places API and partner ecosystem.
Earlier chapters
Twitter · Developer Relations Lead · 2013–2016Apple · Product Manager · 2011–2013Google · Associate Product Manager · 2009–2011