Durable AI workflows with Effect: Maintaining open source at scale
Presented by: Maxwell Brown
Originally aired on October 23 @ 9:00 AM - 9:30 AM EDT
Every project produces signals that help guide its evolution. Products learn from telemetry, usage, and customer behavior, while open source projects like Effect learn from issues, pull requests, discussions, and contributor feedback. The rise of AI-assisted development dramatically increased the volume of these signals, eventually outpacing our ability to review, interpret, and act on them. To address this problem, we built “The Janitor,” an AI maintainer that reviews incoming issues and pull requests, gathers missing context, organizes related work, and surfaces what needs human attention. We show how running durable workflows with Effect on Cloudflare’s Developer Platform helps us separate signal from noise so maintainers can spend time where it matters most.
English
