On November 18, 2026 Google Labs introduced Stitch Loop, an experimental “product agent” that keeps improving software after it ships. Google describes it as a new direction for its Stitch design tool.
What to know
- What it does: Stitch Loop connects to your codebase and live product data, finds issues and opportunities, proposes one or more fixes with tradeoffs (and design options where relevant), writes the code in an isolated cloud sandbox and opens a GitHub pull request for your approval. Google says nothing merges without your review.
- After a change: it watches live data, learns from the results and keeps watching for regressions once a goal is met.
- Loops: three default loops (cybersecurity, latency and accessibility) plus custom loops, such as improving checkout conversion. Google says the security loop combines Gemini 4 Carbon with CodeMender, Google DeepMind’s code-security agent.
- Integrations: GitHub, plus logs, user behaviour and feedback through more than 180 integrations, custom MCP servers or CLI uploads. It can also serve coding agents such as Google Antigravity through the Stitch CLI and MCP, and works in Slack.
- Availability: a waitlist opened on November 18 for users globally, with priority access for eligible Google AI Ultra users. Google does not give a price.
Why it matters
Most coding agents wait for a prompt. Stitch Loop is pitched as one that picks its own work from your metrics, which makes review rules more important, not less. Before you connect any agent to your code and data, read our agent permissions checklist. For the models behind it, see the Gemini 4 model guide and the Gemini page.
Source: Google. We summarise in our own words; check the source for the full details.
Contribute your creativity