Netlify has launched Agent Runners, a new way to start web projects from a text prompt at netlify.new. Pick a leading coding agent — Claude Code, Codex, or Gemini CLI — describe what you want, and get a live web app deployed on Netlify infrastructure in minutes.
The pitch is straightforward: most AI-generated projects hit a wall when you try to take them to production. The prototype works, but then you need to stitch together hosting, functions, auth, and storage from scratch. Netlify's approach skips that gap entirely — the project starts on production infrastructure with platform capabilities already wired in.
What's Available From the Start
- Functions for serverless backend logic
- Identity for authentication
- Blobs for object storage
- Forms for data collection
- AI Gateway for managing model API calls
Code and Prompts on the Same Project
The key design decision: prompt-generated projects and hand-coded projects share the same infrastructure and workflow. Developers can work in code while teammates iterate with prompts on the same project. No rebuild, no migration, no separate toolchain for the AI-generated parts.
Enterprise Governance
A new Internal Builder seat lets product managers, designers, and other non-engineering roles build with agents inside the Netlify organization. Engineering retains oversight of what reaches production — role-based access controls determine who can prototype and who can ship.
This is the hosting platform play for the agent era. Netlify is betting that the entry point for new software shifts from git init to a prompt, and whoever owns that first interaction captures the deployment relationship. Vercel is making a similar move with their coding agent plugin. The race is on to be the default platform agents deploy to.