How Rillet ships 3× faster with AI agents on Vercel

At Rillet, an accounting software company, AI agents turn customer requests into pull requests, and one engineer, Anthony Liang, reviews and merges every change.

Rillet makes accounting software in which AI agents do bookkeeping work for more than 600 customers. Its two-person frontend team lost time gathering customer requests scattered across Linear, Slack and conversations. The team now has one AI agent gather the context and a coding agent draft the change, while an engineer approves each one.

• Customer success staff flag an issue with a browser extension, and an AI agent collects the context from Slack and Linear.
• For safe, well-scoped changes, a coding agent writes a short spec, makes the change and opens a pull request with screenshots from a test browser.
• Anthony Liang, the software engineer who leads the frontend team, reviews and merges each pull request, and the agent resumes its session if he asks for changes.
• Vercel reports that the team closed more than 800 tickets in its first four months; the figures are self-reported and have no stated baseline.

A small team can route customer requests to an AI agent if it limits the agent to well-scoped changes and keeps one named engineer approving every merge. This is an inference from Vercel's customer story about its own products, written by its marketing team.

Derived by Working Surface from the article; more in the Playbook. Source line: Knowing we can control and audit that environment helps me sleep better at night.

11 October 2026: AI agents depend on written context that goes stale, and the teams in these pieces review agent changes more firmly than they keep that context current.