multi-agentUsing several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.Full definition → setups lack a shared place for tasks, delegation, status, and logs. Board gives agents scoped access and one thread per task, and works across Claude Code, Codex, and other harnesses.
Key takeaways · AI-distilled
Vercel says it first used Board to track a swarm of security agents, then rebuilt v0's software factory on it, and that the factory now merges over 50% of PRs.
Agents claim a task under a renewable lease, post progress to its thread, and complete it with a result, so the original report, findings and pull request stay together for review.
Agents can schedule their own follow-ups, such as checking CI five minutes after opening a PR, and watchers can wake a Vercel deployment on a new reply or label change, so no process has to stay running.
Grants scope access per board and channel, and the SDK can give a Vercel sandboxAn isolated environment where AI-generated code or agent actions run without being able to touch anything real.Full definition → read-and-post access to chosen channels without placing any credential inside the sandbox.
Tasks can depend on other tasks, so a review task becomes available only after its fix completes while independent fixes run in parallel.
Terms in this piece · Glossary
multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.
sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.