Zodeen / Live work: the work reports in

Code, builds, tests, databases and AI agents report into the plan by themselves

Zodeen connects the plan to the systems doing the work, so progress shows what is really happening, not just what people remember to type.

The problem

A task says 80% done while its tests fail. Project tools only know what people type into them, and no mainstream tool ties the plan to CI, test suites or AI-worker output.

How it works

Connected systems

Repositories (GitHub), pipelines (GitHub Actions), test suites (Playwright), databases (Postgres), trackers (Jira), chat (Slack) and Claude agents, each tied to the tasks it proves.

A live event stream

Commits, builds, test runs, records migrated and agent runs arrive as they happen.

Isabella acts on them

When tests fail she marks the task at risk and logs the issue with the evidence, without waiting for anyone to type it.

AI workers on the team

Claude agents are team members with capacity, runs, success rate and cost, planned like everyone else.

An example: Brightway’s connections

brightway-app repository: 146 commits this week. Build pipeline: 47 of 49 builds passing.
Backend API test suite: 40 of 46 tests passing, tied to the cloud backend task.
Accounts database: 1.19 million of 1.7 million records migrated; the data migration agent has checked 610,000.
The code review agent has reviewed 14 pull requests.

The value

Progress you can prove.
Problems found by the systems, the day they happen.
AI agents managed like the rest of the team.

In the product

Live work: every connected system, its health, its metric and its tasks.
Failing tests become a logged issue, with the evidence.
AI workers: capacity, runs, success rate and cost.