QAM Hub QAM Hub
Home / Features / AI-assisted testing

AI-assisted testing

AI in QAM Hub proposes and a person decides: test cases and checks generated from a description, a Quality Analyzer that grades a suite and finds duplicates, and autonomous agent runs that drive a real browser through your regression cases and queue the results for review.

AI test case generation

Generate test cases and checklist items from a short description – a QA engineer reviews every result before it is saved.

Describe the feature, paste an acceptance criterion or point at a requirement, and QAM Hub proposes test cases or checklist checks in your project’s own format – using the project AI context, templates and preferences an admin has configured, plus any knowledge documents you have uploaded.

Nothing is written until a human approves it. Results arrive as a reviewable list: edit, drop or accept, then save into the suite. The same applies to AI bug analysis, which turns a messy failure note into a clear reproduction.

  • Test cases and checklist checks from a description or requirement
  • Project-wide AI context, templates and formatting preferences
  • Every proposal reviewed and edited by a human before saving
  • AI bug analysis for turning a failure into a usable report

Quality Analyzer

Finds duplicate, vague and incomplete test cases across a suite, grades it, and tracks what you fixed.

Test bases rot quietly: the same scenario written three times, cases with no expected result, steps that stopped matching the product. The Quality Analyzer reviews a suite in the background, groups likely duplicates, and reports per-case findings with the quoted evidence that triggered each one.

The A–D grade is derived from the distribution of findings rather than asked of a model, so the same suite always scores the same. Findings carry a triage status – fixed, dismissed or still open – and each run is compared with the previous one, so cleanup progress is measurable instead of anecdotal.

  • Duplicate detection, with "not a duplicate" remembered per pair
  • Per-case findings with the quoted evidence behind them
  • A deterministic A–D score and a run-over-run delta
  • Triage: fix with AI, dismiss, or leave open – plus PDF export

Autonomous agent runs

Let an AI agent drive the browser through your regression cases, then triage the results yourself.

Agent flows take a set of test cases and execute them in a real browser through a runner you install with one command. Each case gets a step-by-step trace, a verdict, and – where the run captured one – a baseline to compare future runs against.

Triage stays human. Every not-passed case arrives in a review queue with the agent’s own reasoning: confirm it as a bug, which opens the issue dialog against your tracker; accept a proposed change to the expected result when the product legitimately changed; or reject the finding. A decision always has a visible consequence.

  • One-command runner install, scheduled runs via cron expressions
  • Step-level traces, baselines and per-case verdicts
  • A review queue with filters, comments and CSV export
  • Confirming a bug creates a real issue in your tracker

Project knowledge base

Upload the documents your product actually runs on, and AI features answer in terms of your system, not a generic one.

A project knowledge base holds PDFs, notes, Markdown, CSV and JSON – specs, domain glossaries, environment quirks – with text extracted server-side. Documents can be enabled or disabled individually, and the relevant ones are assembled into a token-budgeted context block for test generation and agent runs.

After each agent run, QAM Hub proposes knowledge updates it learned from what it saw: add this, correct that, this note is stale. People accept or reject them, so the knowledge base stays current without becoming a second documentation chore.

  • PDF, TXT, Markdown, JSON and CSV documents, plus text notes
  • Per-document enable and disable, folders for structure
  • Injected into AI generation and agent runs automatically
  • AI-proposed knowledge updates, accepted or rejected by a human

In-app AI assistant

A product help chat that answers "where is this and how does it work" and links straight to the right screen.

The assistant knows QAM Hub, not the internet. Ask where a setting lives, how run tags differ from case tags, or how to connect a reporter, and it answers with links into your own workspace.

It deliberately refuses to write test cases or bug reports and points you at the built-in AI tools instead – one place to generate QA artifacts, with review built in, rather than two with different rules.

  • Product and navigation answers with in-app deep links
  • Voice input, and an improvement ticket draft when you hit a gap
  • Scoped to the product – it redirects QA-artifact requests
  • Available from every page header

Other areas

Try ai-assisted testing on your own project