# Stitch Loop — Google Labs agent that keeps improving shipped products

> A long-running Google agent that watches your live app, proposes fixes and opens pull requests for review.

- Org: Google
- Date: 2026-11-18
- Category: tool
- Source: https://blog.google/innovation-and-ai/models-and-research/google-labs/stitch-loop/
- Canonical page: https://ai-tldr.dev/releases/google-stitch-loop/

## In short

Stitch Loop is a Google Labs agent that connects to your codebase and live product metrics, suggests fixes, builds them in a cloud sandbox and opens a GitHub pull request. A global waitlist is open.

## What it is

Stitch Loop is an experimental Google Labs agent for software that has already shipped. It reads your codebase and live data such as logs, user behavior and feedback, then suggests improvements backed by evidence from your product. It ships with security, latency and accessibility loops, and you can create custom loops, for example to improve checkout conversion.

## How it works

Each loop runs four steps: suggestions, solutions with tradeoffs and design options, a build step that writes code in an isolated cloud sandbox and opens a GitHub pull request, and a reflect step that tracks the live metric afterwards. Nothing merges until a person reviews the PR. The security loop pairs Gemini 4 Carbon with Google DeepMind's CodeMender, and inputs arrive through 180+ integrations, custom MCP servers or local CLI uploads.

## Why it matters

Most coding agents stop when the code is written. Google is aiming this one at the work after launch, where teams watch metrics and fix regressions. It also works as a service layer for agents like Google Antigravity through the Stitch CLI and MCP, and posts daily digests in Slack via @stitch.

## Who it's for

product and engineering teams with live apps

## Try it

https://stitch.google.com/loop/waitlist

## Links

- [Announcement](https://blog.google/innovation-and-ai/models-and-research/google-labs/stitch-loop/)
- [Waitlist](https://stitch.google.com/loop/waitlist)
- [CodeMender](https://deepmind.google/blog/introducing-codemender-an-ai-agent-for-code-security/)
- [Gemini 4 Carbon](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-carbon-now-available/)
