Hi, I'm Matt.

Product Manager at Run.Game. I build creator tools, and the AI agents that help ship them.

Previously led VALORANT Premier at Riot. Based in Los Angeles.

Selected work

A Slack-native Claude agent that triages inbound, files Jira tickets, and learns across sessions.

Riot's always-on competitive tournament system for VALORANT. Shipped globally.

Web3 4X grand strategy game. 2 years of production as General Manager.

Western-themed idle game. Sole product owner from concept through live ops at Kongregate.

May 2026

Shipping the next iteration of Run.Game’s Creator Studio. Better project scaffolding, faster iteration loops.
Exploring how to push our internal agents toward more async, autonomous work. What “set and forget” actually looks like in practice.
Writing one long-form post per month. Next one is on how we can make collaborating between multiple humans and multiple agents better.
Experimenting with runtime generative AI in games. Figuring out which surfaces are worth generating at play time vs. baked in.
Reading: The Formula (Robinson & Clegg), Delivered from Distraction (Hallowell), Turn the Ship Around (Marquet)

A year of building AI agents, four rules of thumb I keep coming back to, and the one that ties the other three together.

Most agent-design advice tells me to give my agents personas. The research says don't. Or at least, not the way most agent tooling assumes.

Every agent I build hits the same fork in the road. New capability needed. Should it run in the agent's context or in its own?