Lately I’ve been thinking about how much QA effort still goes into maintaining the testing process itself instead of actually improving product quality.
Not test execution or automation — the process around it. Updating test cases nobody really trusts anymore, re-linking traceability after changes, explaining coverage in meetings, running massive regression suites because nobody feels comfortable removing anything.
What surprised me is that AI hasn’t really changed the hardest parts of QA for my team. Generating test cases is easy now. Generating confidence is still hard.
The teams I see moving faster usually aren’t the ones with the biggest automation coverage. They’re the ones that are better at understanding what actually changed, what’s truly at risk in a release, and what really needs validation right now versus what’s just there because it’s always been there.
I also think a lot of QA organizations are quietly moving away from the idea that more tests automatically mean better quality. In large products especially, too much testing can create its own problems — more maintenance, more noise, slower decisions, and less trust in the results.
Curious whether others here are seeing the same thing. Has your biggest QA bottleneck changed over the last year, or is it still mostly execution and automation?