Netlify AI: Agent Runners and AI Gateway Explained

Netlify AI is a set of AI features built into the Netlify platform that let you build, run, and ship apps with AI agents. The two headline tools are Agent Runners, which turn a prompt into a deployed app, and the AI Gateway, which gives your app access to AI models without managing API keys.

This guide explains what each Netlify AI feature does, how to use it, and what it costs. We keep the language plain and stick to what Netlify's own product pages describe.

Netlify prices usage with credits, and AI inference uses those credits, so plan choice matters. Confirm the live limits on Netlify's pricing page before you rely on them.

What is Netlify AI?

Netlify AI is Netlify's built-in tooling for building and running apps with AI agents. It sits inside the same platform you already use to deploy sites, so the AI features share your deploys, previews, and hosting.

There are two main pieces. Agent Runners let an AI agent build or change an app for you. The AI Gateway lets the app you ship call AI models without you wiring up API keys.

The goal is to remove setup friction. You describe what you want or add AI to your app, and Netlify handles the plumbing around it.

Netlify Agent Runners, explained

Agent Runners let you spin up a new project from a prompt or iterate directly on your live app. An AI coding agent makes the change, and Netlify deploys a preview of it automatically.

You can use major models with it. Netlify lists support for Claude, Gemini, and Codex, so you pick the agent you trust for the job.

The design removes the usual setup. There are no API keys to manage, every change gets its own Deploy Preview, and the whole thing plugs into approval workflows so a human signs off before it ships.

• Start from a prompt, or point the agent at your existing live app to iterate.
• Works with Claude, Gemini, and Codex agents.
• Each change gets an automatic Deploy Preview to review before merging.
• No API keys or manual setup to get started.
• Approval workflows keep a human in the loop before production.

Netlify AI Gateway, explained

The AI Gateway gives your app access to any AI model with zero API keys. Instead of storing provider keys and wiring each model yourself, your app calls models through Netlify.

This matters for two reasons. It removes the risk of leaking API keys in your code, and it lets you switch or add models without re-plumbing your app.

For a small team shipping an AI feature, that is real time saved. You focus on the feature, not on secret management and provider setup.

How to use Netlify AI

You use Netlify AI from inside a Netlify project, so the first step is connecting a site or starting a new one. From there, the AI tools are a few clicks, not a separate service to learn.

Start with Agent Runners to build or change an app, then use the AI Gateway to add model calls to that app. Review each Deploy Preview before you ship.

A simple first run looks like this.

• Create a Netlify project, or connect an existing repo.
• Use Agent Runners: give a prompt to build, or point the agent at your live app to change it.
• Open the auto-generated Deploy Preview and review the change.
• Add AI model calls in your app through the AI Gateway, no keys required.
• Approve the change so it deploys to production.
• Watch your credit usage, since AI inference draws on your plan's credits.

Netlify AI pricing: what it costs

Netlify AI has no separate price — it uses the credits included in your Netlify plan. AI inference, builds, and bandwidth all draw from the same credit pool, so a heavier AI app uses more credits.

Plans run from Free to Enterprise, and the credit allowance rises with each tier. The table shows the plan, its price, and its included credits.

Because AI inference spends credits, pick your plan by how much you expect to build and run. Free is fine to test; a live AI app usually needs Personal or Pro.

Is Netlify AI worth it?

Netlify AI is worth it for developers and small teams who already deploy on Netlify and want to build or add AI features fast. Because the AI tools share your existing deploys and previews, there is little extra to set up.

It is a strong fit if you value speed and simplicity. Agent Runners remove the setup for AI coding, and the AI Gateway removes API-key management, so you ship sooner.

It is less of a fit if you need deep control over model routing, self-hosted infrastructure, or a provider-agnostic setup outside a single platform. In that case a dedicated AI infrastructure stack may serve you better.