Vercel Board gives coding agents a shared task history

Vercel announced Board on 18 November, an experimental message store for coordinating people and coding agents. It gives work a shared thread rather than leaving the original request, investigation…

SSophie Harrington1 min read

Vercel announced Board on 18 November, an experimental message store for coordinating people and coding agents. It gives work a shared thread rather than leaving the original request, investigation and result scattered across separate agent sessions.

For a team maintaining a commerce storefront or Shopify app, the potential use is development coordination. Board is not a merchant-facing Shopify integration, and connecting it does not grant an agent permission to deploy a storefront.

Keep the task and its evidence together

Agents access Board through a CLI or JavaScript SDK. Channels and threads hold discussions, while labels can identify status or a related pull request. Agents can claim tasks, report progress and record results. Access grants limit which channels a person or agent can read or write.

The release notice says Board is available on all plans and describes two-way message synchronisation with Slack, GitHub and other tools through Vercel Connect. Teams can use their own agents and execution environments. Availability on a plan should not be read as a promise that every connected service or compute workload is free.

A small storefront test

Our suggested trial is a single non-production bug: an incorrect product-card layout on a preview deployment. Keep the reported behaviour, screenshots, proposed fix, test result and reviewer decision in one thread. Give the agent only the access needed for that task, and leave production deployment behind the team’s existing approval process.

The useful result is traceability: another developer should be able to see what was requested, what changed and which checks passed. A completed task label alone does not prove that checkout, accessibility or pricing still works.

Vercel’s announcement includes an illustrative Board demonstration. Its caption says no models ran and the people, agents and results are illustrative. Treat it as a walkthrough of coordination mechanics, not independent evidence of agent reliability or merchant productivity gains.

ℹ️
Board is experimental coordination infrastructure. It does not itself approve a code change or establish that an agent’s work is correct.
Filed underAI

More from AI

See all →