Introducing Tembo Automations
Introducing Tembo Automations: a new primitive for background agents that lets you automate away repetitive tasks without you even having to think.
If you prefer tweet format, read the launch announcement on X.
How It Works
Automations can run on a schedule, react to any webhook, or activate just by mentioning @Tembo. Everything runs inside a sandboxed VM, so agents can execute code, run full apps, call CLI tools, and use any MCP server you expose.
Tembo always has full organizational context:
- Your codebase
- Your documents
- Your integrations
- Your MCP servers
This is what allows automations to make real decisions based on how your system actually works, not a simplified version of it.
Example Use Cases
Below are a few examples teams are already using every day.
PR Reviews
Spin up a VM with any coding agent and let it review the PR. Configure the setup per engineer, per team, or apply one global configuration. High-quality reviews, fully automated.
Fix Sentry Errors
Listen to Sentry webhooks, generate a fix, open a PR, and notify Slack. Tembo turns error alerts into actionable patches.
Enrich Linear Issues
Tembo adds comments with deeper context:
- Relevant files
- Integration data
- Draft solutions
- Hints and guidance
Issues become clearer, more actionable, and easier to estimate.
Auto-Update Documentation
Push a new release or refactor a repo and Tembo keeps your docs up to date in the right place. No more stale folders drifting out of sync.
Send Merged PRs to Slack
Simple but useful: listen to the PR merged trigger and keep the team informed in real time.
Design Your First Automation
Start with one repetitive task and define what a successful run should produce. For a Sentry alert, that might be a pull request with a focused fix and a description of how it was tested. For a Linear issue, it might be a comment that identifies relevant files and unanswered questions, without changing any code.
Write the instructions around that output. Specify the repository, the context to inspect, the checks to run, and where the result should go. Be explicit about what the agent should leave alone and when it should ask for a human decision instead of proceeding.
Try the workflow on a representative event before relying on it for a busy queue. Review whether the result was useful, whether the scope stayed narrow, and whether the notification reached the right people. Refine those instructions before adding more triggers or broader responsibilities.
Build Whatever You Need
And this is just a small slice. Once you combine triggers, MCP, and sandboxed execution, you can build whatever automation your team needs: from daily security scans to personalized PR reviewers.
Tembo Automations is live now.
We can't wait to see what you build.
Run any coding agent in the cloud
Tembo agents execute tasks in secure cloud environments and return reviewable output. Use any agent or model, run in parallel, and keep humans in control.
