Vercel Agent

Vercel Agent lives in your dashboard and can investigate what's happening in production, answer questions about your projects, and take action on your behalf.

Because Agent is built into Vercel's platform that deploys and serves your app, it can read the signals around it: deployments, logs, metrics, project configuration, usage, and connected repositories. That context is what turns a question into an answer and a problem into a fix.

Vercel Agent runs on Vercel's AI Cloud. It can use secure sandboxes to reproduce issues, validate generated code, and run checks before suggested changes reach production.

Dashboard chat, investigations, and approved actions are in public beta for Pro and Enterprise teams. Rollout will be gradual. If you don't have access yet, you can request access.

Get automatic code reviews on every pull request. Code Review gives you Sandbox-validated suggestions on your pull requests as one capability within the broader Vercel Agent.

What it does:

• Performs multi-step reasoning to identify security vulnerabilities, logic errors, and performance issues
• Generates patches and runs them in secure sandboxes with your real builds, tests, and linters
• Only suggests fixes that pass validation checks, allowing you to apply specific code changes with one click

You can also mention @vercel in any pull request comment. The agent will read your message and either propose a fix for you to review and apply, or respond directly to your question in the same thread.

Learn more in the Code Review docs.

When anomaly alerts fire, Vercel Agent Investigations can analyze what is happening in production. Point Agent at a failed deploy, a runtime error, or a cost spike, and it traces the cause and recommends a fix.

What it does:

• Queries logs and metrics around the time of the alert
• Looks for patterns and correlations that might explain the problem
• Provides insights about potential root causes

Learn more in the Agent Investigation docs.

When a task requires write access, Vercel Agent presents a scoped plan and waits for your approval. With your sign-off, Agent can open a pull request, roll back, or update a config to remediate an issue.

Agent is read-only by default and cannot make changes until you approve the plan.

Add Web Analytics and Speed Insights to your project using Vercel Agent. Instead of manually installing and writing integration code, Vercel Agent analyzes your repository, installs dependencies, writes integration code, and creates a pull request. All you need to do is review and merge.

Learn more in the Agent Installation docs.

Vercel Agent runs under its own identity and is bounded by the requesting user's permissions. It is read-only by default.

When Agent needs elevated access, it requests a scoped plan and makes no changes until you approve it. Generated code runs in Vercel Sandbox before it reaches production, and elevated actions are attributed to Agent, the requester, and the approver.

You can enable Vercel Agent in the Agent section in the sidebar of your dashboard. Setup varies by feature:

• Code Review: You'll need to configure which repositories to review and whether to review draft PRs. See Code Review setup for details.
• Agent Investigation: This requires Observability Plus and in order to run investigations automatically, you'll need to enable Vercel Agent Investigations. See Investigation setup to get started.
• Installation: See Installation docs for details.

Simple requests, like finding a setting, explaining a DNS warning, or answering a docs question, are free for a limited number of messages during the beta.

Requests that require Agent to investigate, plan, or write code are billed on demand: provider tokens at the underlying rate with no markup, plus a Vercel Token Rate of $0.25 per million tokens.

This rate also applies to Agent Code Review.

Vercel Agent pricing depends on the feature:

• Chat: Simple requests are free for a limited number of messages during the beta. Paid Chat uses provider tokens at the underlying rate with no markup, plus the Vercel Token Rate.
• Code Review: Billed on demand: provider tokens at the underlying rate with no markup, plus the Vercel Token Rate. This is the same pricing as paid Chat.
• Agent Investigation: Includes 10 alert investigations per billing cycle with Observability Plus. Additional investigations use the same token-based pricing as paid Chat and Code Review.
• Installation: Agent Installation is free for all teams.

Existing Agent credit balances are used first. After existing credit is consumed, additional paid usage is billed on-demand once usage billing is enabled. For complete pricing details, see Vercel Agent Pricing.

Vercel Agent never trains on customer code if your Vercel team's data preferences setting is "off" or you are on an Enterprise plan.