AI has made contributing to open source dramatically easier. For maintainers, that also means more issues, more pull requests, and more code to evaluate without any increase in maintainer time.

I work on Laravel's open source team across projects including Boost, Laravel AI, MCP, and Roster, which together have accumulated more than 100 million package installs and nearly 7,000+ GitHub stars. On a simple days, I may work through 10 to 20 community contributions alongside my own engineering work.

The challenge is not simply reviewing contributions faster. Laravel is intentionally opinionated about APIs, conventions, backwards compatibility, and developer experience, so technically correct code may still not belong in the project.

In this talk, I'll show the AI-assisted workflow I use to scale that work: the harnesses and repository context I give coding agents, the guardrails that keep automation safe, and how I use agents to investigate issues, reproduce bugs, trace code, run tests, check edge cases, and prepare first-pass reviews.

The final merge decision remains human.

The goal is to automate the repetitive work without automating away the judgment, project direction, and taste that make an open source project distinctive.