Skip to content
Agents tracked: 501 Releases (7d): 777 Agent status: no reported issuesSince last limit reset: Claude 10d · Codex 8d For AI agents: Markdown Updated Nov 21, 2026
Case study RilletResults as reported by Vercel

How Rillet ships customer-requested changes faster with eve agents on Vercel

Rillet, an AI-native ERP vendor, standardized its internal agents on Vercel's eve framework and runs them on Vercel. A two-person frontend team now turns customer success requests into reviewed pull requests, with Slack and Linear access and model routing handled through Vercel services.

The problem

When a customer asked for a change, the request was spread across a Linear ticket, a Slack thread and a follow-up conversation with an engineer. By the time anyone had full context, writing the code was the quickest part of the job. Rillet also wanted agent-building conventions that people who had never used TypeScript could follow.

How they did it

  1. Define each agent as an eve directory holding instructions in markdown, tools in TypeScript, skills, subagents, channels and schedules.
  2. Deploy all agents into one company-controlled Vercel organization.
  3. Use Vercel Connect to give agents reusable, short-lived-token access to Slack and Linear.
  4. Route model requests to Anthropic and OpenAI through AI Gateway, choosing the model with a string in the agent configuration.
  5. Have customer success describe a change in a Chrome extension, whose eve agent captures context and hands safe, well-scoped changes to a coding agent that opens a pull request.
  6. Let agents test changes in a sandboxed browser and attach screenshots, then have an engineer review and merge, with the agent resuming its session if changes are requested.
  7. Manage builder access with Enterprise Managed Users, Google Workspace SSO and Directory Sync.

From the official docs

import { defineAgent } from "eve";

export default defineAgent({
  model: "zai/glm-5.3",
});

Results as reported by Vercel

  • As reported by Vercel, Rillet tripled its shipping rate in three months.
  • As reported by Vercel, the two-person team closed 800+ Linear tickets in its first 4 months.
  • As reported by Vercel, customer-requested changes reach production in as little as two hours, via Rillet's in-house agent Nebula.
  • As reported by Vercel, Rillet's CEO said the company shipped more in a few weeks than in the whole first half of 2026.

These figures are as reported by Vercel in Vercel customer story ↗ (Oct 9, 2026). Vercel sells the product used here. AgentGid did not measure or independently verify them.

Takeaway. Putting each agent in a standard directory and letting the person closest to the customer start the change can shorten the path from request to reviewed pull request.
AgentGid's take

Best for a small team already on Vercel, with someone who can write TypeScript tools and set up Slack, Linear and SSO access. Eve is free and open source but still in beta, and it deploys to Vercel, whose hosting is priced separately. Every result here is reported by Vercel, and engineer review of each pull request remains. AI SDK (Free (OSS)) is a growing alternative if you want less lock-in.

The agent used here

Similar use cases