Developers are bringing ideas online faster than ever before, thanks to AI.
AI agents like Claude Code and Gemini CLI can write amazing code. Some can even deploy it. But they’re missing crucial context—they can’t see your production logs, don’t understand your build configuration, and have no idea why that last deploy failed.
Meanwhile, when your designer wants to tweak styles or your PM spots a typo, they’re still sending you Slack messages. The AI tools that could help them contribute directly are trapped in a Docker container or behind a local development setup.
What if, instead, AI agents had the full context of your production environment? What if your entire team could use them instantly from anywhere without any setup?
That’s what we built with Agent Runners.
Prompt AI with full production context
Agent Runners brings AI coding agents directly to your Netlify project. Available from the Netlify Dashboard, Claude Code, Gemini CLI, and Codex run in isolated sandboxes with full context of your production environment with no local setup required.
Here’s how it works:
- Open the Netlify Dashboard (or trigger via CLI)
- Choose your AI agent and describe what you need
- Changes create a shareable deploy preview
- Review and merge when ready
No more copying error messages or explaining your setup. The agent understands your codebase, your production environment, your platform. You describe what you want, the agent understands your current state, and changes flow through your deployment pipeline. Just as if the agent is an extension of your team.
Moving fast, and not breaking things
We designed Agent Runners to give developers the right level of control. AI agents get powerful access to ship real work, while you maintain complete oversight:
- Isolated execution sandboxes. Agents run in isolated environments. Nothing touches production until you approve.
- Human-in-the-loop deployment. Every change creates a deploy preview, pull request, and full diffs.
- Secrets stay secret. Agents see only what they need. API keys remain hidden, and updates are scanned before deploying.
- Full audit trail. Every run is logged: who initiated it, what was prompted, what changed, who approved it.
Your existing review process stays intact. Just accelerated development and saved capacity.
What teams are actually building
The real test of any development tool is what people create with it. Here’s a glimpse of what happened when we first got this into our user’s hands.
Fixing errors faster
Netlify already tells you why builds fail with AI-generated summaries. Now Agent Runners takes it further—pull that context directly into an agent run and get a PR fix in clicks. No more scrolling through endless build logs or asking ChatGPT.
Clearing the backlog
Everyone has a list of “someday” improvements. Small fixes that never justify pulling a developer off feature work. With Agent Runners, teams are finally tackling these backlogs. The agent understands your production context, so you can describe the fix you want and let it handle the implementation while you go about your day. Those accessibility improvements, the UI polish, the performance optimizations that have been waiting for months—they’re finally getting done.
Working from anywhere
Ben Robertson, founder of Colorbliss, an AI coloring book generator, shared a moment that resonates with me and captures the shift:
“I had a little update that I forgot to make while I was in the office on Friday to get a new page live. I was able to use Agent Runners on my phone to get the change added and pushed live while I was getting dinner ready for the kids. Very nice!”
No laptop, no environment setup, just describe, deploy, and go.
Scaling changes across projects
Balki Raman, CTO of Vinomofo, is using Agent Runners for everything from rebranding to rethinking content strategy. His team recently updated favicons and logos across multiple sites—what could have been a tedious manual process became a single conversation with an agent.
They’re also exploring how to make their storefront more AEO (Answer Engine Optimization) friendly—critical as AI models change how they surface content, but an expensive slog to maintain manually across hundreds of pages.
Democratizing code contributions
Here at Netlify, designers tweak components, marketers update copy, PMs fix bugs—all without local setup. They describe changes, Agent Runners handles the code, everything flows through standard review.
What’s ahead for Agent Runners
Agent Runners today is just the beginning. Coming soon:
- Webhooks – Trigger agents automatically from internal and external events. Build fails? Customer reports bug? Agent starts fixing immediately.
- Ask mode – Chat about your codebase without making changes. Understand dependencies, explore refactoring options, get architecture recommendations.
- Slack integration – Prompt Agent Runners directly from Slack. See run updates and results where your team already works.
- MCP support – Connect agents to third-party context through Model Context Protocol. Pull in data from your other tools and services.
- Collaborative previews – Deploy preview comments become context and triggers.
Now, AI agents can become part of your team, understanding your infrastructure and shipping real work. Not replacing developers, but amplifying and enabling everyone to become one.
Get started today
Agent Runners is available now on all credit-based plans. Start your first run with Agent Runners in the Netlify Dashboard.
Want to see what others are building? Join the NTL DEPLOY challenge and share what you create with Agent Runners for a chance to win prizes. The convergence of AI development and infrastructure is here. Your next PR is just a prompt away.
