CI/CD that runs itself, and doesn't run up the bill.
CICaDo is CI/CD built for the age of AI agents. Your agents push the code and fix what breaks. CICaDo runs every job on fast, isolated machines, without the servers, plugins and people your old CI needs. You check in from chat when a human needs to decide.
Keep your team out of the tooling
CI should be something that happens to your code, not a second job for your engineers.
Costs less
1,000 free minutes a month, then $0.006 a minute. Priced per organisation, not per seat, so adding people or agents costs nothing. No build servers to pay for or look after.
Gets out of the way
Every job starts on a fresh, isolated machine and runs in parallel wherever it can. No queues behind a shared build box, no plugins to update, no agents to patch.
Your agents do the work
Coding agents connect over MCP, read why a run failed, push the fix and watch the next run, without anyone opening a dashboard.
You check in from chat
Releases wait for a person when they should. Approve, reject or look at the run from Slack, and every decision is recorded.
The new loop: agents build, you decide
1. Every change runs, in parallel
A push from a person or an agent starts the pipeline from a YAML file in the repository. Independent jobs run side by side on their own machines, so a run takes as long as its slowest path, not the sum of its jobs.
2. Your agent reads the failure and fixes it
CICaDo's MCP server gives coding agents the run status and the exact failing lines, with secrets already masked. The agent pushes a fix and the run starts again. Nobody had to dig through a log.
3. You approve the release from chat
When the tests pass and a deploy needs a person, the right people get a message with the commit and results. One click ships it. That is the only part of the pipeline that needs you.
Bring your legacy CI into the age of AI
Jenkins and the build servers like it were made for a world where people watched builds. They need someone to patch the controller, keep plugins compatible and babysit build agents, and they keep their results where an AI agent cannot reach them.
CICaDo is a full CI/CD system that plugs into the tools you already have. Your pipelines move into your repository as YAML, run on hosted machines, and become something your agents can read and act on.
# cicado-workflow
name: ci
on:
push:
branches: [main]
pull_request: {}
jobs:
test:
runs-on: [linux]
steps:
- name: test
run: make test
deploy:
needs: [test]
runs-on: [linux]
steps:
- name: deploy
run: ./deploy.sh
Connects to your tools
| Area | Available in the beta | Next |
|---|---|---|
| Source hosts | GitHub | GitLab, Bitbucket and plain Git |
| Chat and approvals | Slack | Microsoft Teams |
| Cloud access | AWS, with short-lived tokens | Google Cloud, Azure |
| AI agents | Any MCP-capable agent | |
| Runners | Hosted Linux runners | Your own runners on paid plans |
Everything else a pipeline needs
Pipelines as code
Jobs, dependencies, schedules and approval steps in one YAML file. Mistakes are reported with the line and column before anything runs.
Results where you work
Every commit and change request gets a status with a link to live logs, reported back to your source host.
Secrets and cloud access
Secrets are encrypted, handed only to the job that needs them, and masked in logs. Jobs sign in to your cloud with short-lived tokens instead of stored keys.
Your own runners too
On paid plans, connect your own machines for special hardware or private networks, alongside the hosted ones.
Stop paying people to babysit builds
We are opening CICaDo to a small number of teams first. Tell us where your code lives and what you build today, and we will send an invite.
Prefer to run CI on your own servers? Self-hosted editions are planned after the hosted launch.