# Test Automation

> A scoped AI Pod that converts manual regression testing into automated gates the pipeline enforces on every merge. 80 to 200 story points.

Source: https://www.koombea.com/ai-pods/catalog/test-automation/

---

Manual regression testing is the reason your releases are monthly. This Pod converts it into scenarios the pipeline enforces, so shipping stops requiring a person and a spreadsheet.


## The problem

Manual regression sets the ceiling on how often you can ship. Verifying a release costs two days, so you release monthly. Every release then carries a month of changes. Large releases fail more often, which makes everyone more cautious, which makes releases larger.

Teams break that loop by writing tests. Then they break the tests by not running them on every merge. A suite that runs nightly and is red most mornings is not a safety net. It is a notification everyone muted.

So we build gates, not reports. A scenario that does not block a merge does not change behavior.

- Every scenario runs on every merge, not nightly
- A red gate blocks the merge, with no override by convention
- Flaky tests are treated as defects and fixed, not retried

## What this Pod ships

- **The critical journeys first**: The paths that carry revenue, automated end to end, because a suite that covers everything equally covers the important things too late.
- **Test data that is not a person**: Fixtures and seeding, so running the suite does not depend on someone remembering to reset an environment.
- **Pipeline gates**: Wired into CI as a required check, with the run time kept short enough that nobody is tempted to skip it.
- **Coverage you can read**: Reporting on what is covered and what is not, so the remaining risk is a known list rather than an assumption.

## Why the order matters

Scenarios come from the spec, before the implementation they verify.

On an existing product there is no spec to write from. So the first step is establishing what the current behavior actually is, and confirming it with you. That conversation routinely surfaces two or three behaviors nobody intended and everybody now depends on.


## Where the suite sits


## What it costs

80 to 200 story points. The drivers are the number of critical journeys and how testable the application is today. An application with seams and a stable interface is quick. One where everything is welded to the interface needs untangling first.

We scope from a walkthrough of your current release process. The estimate then reflects the suite you need, not a coverage percentage someone picked.

- Points: 80 to 200, committed before work begins
- Scoped from your release process, not from a coverage target
- Playwright, Cypress, and the frameworks your team already runs
- Handover includes how to write the next one yourselves