Manual QA + AI + Playwright: who does what?

Two questions that show up in every planning meeting:

1. "AI can test everything now — do we still need manual QA?"
2. "We have Playwright — why click through manually?"

Short answer: all three do different jobs. Mixing roles is how you pay for automation and still miss subtle bugs.

Division of labor in one line

Does AI replace QA?

No — if you need someone accountable for product quality.

AI is strong at:

• test ideas from user stories
• bug reports from rough notes
• turning huge logs into timelines

AI lacks:

• business context ("this flow is 40% of revenue")
• UX feel ("this lag will make people bounce")
• accountability for Go/No-Go

Treat AI like a fast intern: draft, human review, human judgment ships.

Does Playwright replace manual testing?

No — it replaces well-understood repetition.

Manual QA should spend time on:

• exploratory on fresh builds
• undocumented edge cases
• real devices and a11y spot checks

Playwright should guard:

• login → core flow → expected outcome
• every painful production bug you've seen
• 3–5 minute smoke after each deploy

See Playwright in real software teams for automation depth, and Manual QA + AI across the lifecycle for AI templates per phase.

One sprint — who does what? (diagram)

Sprint week (small team)

Planning
  PM/BA: stories + AC
  QA + AI: draft edge cases → QA picks real risks
  Dev: estimate

Development (parallel)
  Dev: code + unit tests
  QA: exploratory on preview branch
  AI: draft exploratory charter (QA edits)

Before merge
  Dev: PR + change summary
  QA: manual smoke on new path
  CI: Playwright critical path

Pre-release
  QA: regression focus (AI suggests scope from changelog)
  Playwright: full smoke + bug-regression suite
  QA: Go / No-Go (human, not bot)

Post-deploy
  Playwright: prod smoke (safe data)
  QA: spot-check + watch support
  AI: summarize week's bugs → retro

Print it. The point is one glance shows ownership.

Concrete example: "change email"

Common mistakes

1. "Let AI run E2E" — not reliable as a production gate; use AI for design/reporting, not instead of Playwright.
2. "100% automation" — suite bloats, gets flaky, team stops running it.
3. "QA only follows scripts" — exploratory skill atrophies; weird bugs slip through.

Practical conclusion

• AI = faster drafts
• QA = quality ownership
• Playwright = regression memory

Combine all three for speed and trust.

---

Share the sprint diagram with your team — look for gaps where nobody owns a step.