Netlify

Filed by the Vibe Code Weekly desk · Updated Nov 12, 2026

The story

Netlify presents itself as a full application platform pitched at agent-driven development: coding agents can scaffold and edit projects inside it, and the resulting code lands on the same edge infrastructure, preview system and backend services that hand-written code uses.

The interesting shift at Netlify is where AI sits in the workflow. Its Agent Runners feature lets Claude, Gemini, Codex or OpenCode generate a new project from a prompt or edit a live app, and Netlify says this requires no setup or API keys. The homepage also invites developers to point their own coding agent at netlify.ai. Traditional routes remain: auto-deploys from Git, the netlify-cli package for terminal deploys, and Netlify Drop for uploading a folder. Whichever route is chosen, the output becomes one project on one set of infrastructure.

That matters because AI-generated code still needs review. Netlify routes every branch, pull request and agent run to its own Deploy Preview URL, which can be shared with teammates, clients or stakeholders and promoted to production in one click once approved. Rollbacks are also one click. For teams letting non-engineers prompt changes, such as marketers editing pages, this preview-then-approve loop is the safety layer, and Netlify says Agent Runners fit existing approval workflows.

Behind the frontend, Netlify bundles what it calls primitives: serverless, edge, background and scheduled functions; Netlify Database for managed Postgres; Netlify Blobs for files; Identity for login; Forms that collect submissions with a data-netlify attribute; an Image CDN; and AI Gateway for calling AI models. Delivery runs on an edge network of 100+ locations with HTTPS, DDoS protection and auto-scaling on by default. Enterprise buyers get SSO/SAML, role-based access, audit logs, a 99.99% uptime SLA and 24×7×365 support.

Netlify is pitching itself as a platform for agent-driven development. Its Agent Runners let Claude, Gemini, Codex or OpenCode start a project from a prompt or edit a live app, and the output lands on the same infrastructure as hand-written code.

The good

Built-in agent support for Claude, Gemini, Codex and OpenCode. Preview URL generated for each branch, PR and agent run. Database, blob storage, auth and forms included in the same platform. Anonymous CLI deploys and folder uploads for quick starts. Enterprise controls including SSO/SAML, audit logs and an uptime SLA.

The catch

Hosting is tied to Netlify's own edge network. Backend logic follows a serverless and edge function model. Enterprise capabilities are sold through a demo request rather than self-serve.

Who should read on

Teams that let AI agents write code but want human review before anything ships. Frontend developers who need APIs, a Postgres database and file storage without separate vendors. Marketing teams publishing campaign pages that need image resizing and form capture. Product teams prototyping AI features that call several model providers.

Who can skip it

Organizations that must run workloads on their own cloud accounts. Applications built around always-on, stateful server processes.

Reader mail