Repository
Suite code in your organisation, with a README to run it locally and in CI.
Service 02 · Automation
Test automation
We automate what is already stable and will be repeated. We start with the cases that stop a release.
Automation does not replace a tester’s judgement. It replaces repetition. Anything still exploratory, ambiguous, or one-off stays manual — and we put that in writing.
01What we automate
What we automate
- Smoke + release-critical regression first.
- Web, mobile and application testing, across internal and third-party systems.
- CI: the suite should fail the pipeline, not be a decorative check.
- Agreed upkeep. If there are no hours for it, the suite dies; we say so before starting.
What we don’t
- Replace a tester’s judgement on exploratory work.
- Leave a flaky test in the suite. A run the team stops trusting is worse than no run at all.
- Run a suite against production data. Test environments, with masked or synthetic data.
- Hold the suite: it lives in your repo, under your licence.
If you already use AI case-generation or selector tools, we treat them like any other tool: keep them if they cut upkeep, drop them if they add it.
Independent testing and tool sales are two separate lines. In a QA assessment we do not recommend a tool because we sell it, and if there is a commercial interest in the recommendation, we say so in writing before you decide.
02How we start
From manual suite to pipeline
Four steps. Exact days depend on the product.
01
Inventory and rule
Which cases exist, which are stable, and which repeat every release. The output is a short list, not a backlog.
02
Pilot
The first critical cases automated and running. It measures real upkeep before you commit to the rest.
03
CI
The suite enters the pipeline with a threshold that blocks the merge.
04
Handover
Repo, instructions to run it, and agreed upkeep hours. Your team runs it without us.
03Deliverables
Automation rule
A short document: which case enters the suite, which stays manual, and why.
Run report
Result per case, real failures versus flaky, and what upkeep the next week needs.
Let’s talk about a pilot.
Ten automated critical cases say more than a 400-case coverage plan.