TR Book a 30-min call

Resources · Tutorials

See AI Testing Probe at work, step by step

This page follows one example Jira story through the six stages, from requirement to a reviewed Playwright test in your repo. Each step shows who acts: AI drafts, your experts approve and the system runs the checks. Every step links to the documentation section with the details.

See Testing Probe in action

An illustrative walkthrough of one example story, STORY-142 Order cancellation, through the six stages.

  1. Stage 1: Requirement intake

    A story is pulled from Jira read-only, with its description and acceptance criteria.

    Who acts:SystemHuman

    Read more in the documentation

    Illustrative
    1. System Jira, read-only STORY-142 pulled · 4 acceptance criteria Done
    2. Human Analyst scope AC-1 … AC-4 selected for this run Done
    3. System Context attached description + acceptance criteria Done

  2. Stage 2: Pre-review and consistency check

    AI compares the story with the app's code and data; each finding carries re-verified evidence and the product owner approves in Jira.

    Who acts:AIHuman

    Read more in the documentation

    Illustrative
    1. AI Consistency check 3 findings · evidence re-verified Done
    2. AI Finding F-2 AC-3 vs code: cancel window differs Done
    3. Human Analyst decision 2 accepted · 1 rejected Done
    4. Human PO approval in Jira label added by the PO · gate open Done

  3. Stage 3: Test cases

    AI drafts cases tied to acceptance criteria; the analyst reviews each one and publishes approved cases to Jira or Xray.

    Who acts:AIHuman

    Read more in the documentation

    Illustrative
    1. AI Case drafts 8 drafts tied to AC-1 … AC-4 Done
    2. Human Analyst review 7 approved · 1 rejected Done
    3. System Publish to Xray 7 approved cases · preview confirmed Done

  4. Stage 4: Playwright code and Break-to-Verify

    An agent verifies each step in the running app, code is filled into a template, and Break-to-Verify shows the test can fail.

    Who acts:AISystemHuman

    Read more in the documentation

    Illustrative
    1. AI Step verification each step checked in the running app Done
    2. AI Playwright code PR #27 · template-bound · page objects Done
    3. System Break-to-Verify green → red on broken build → green Done
    4. Human Merge approval engineer review Done

  5. Stage 5: CI and Control Tower

    Approved tests run on PR, push, deploy and nightly triggers with no AI in CI; results reach the Control Tower.

    Who acts:System

    Read more in the documentation

    Illustrative
    1. System Triggers PR · push · deploy · nightly Done
    2. System Nightly CI 42/42 passed · 0 AI calls Done
    3. System Control Tower results with their source Done

  6. Stage 6: Triage, repair and bug drafts

    Rules classify failures before AI; a locator break gets a one-line repair PR, a real defect gets a bug draft for the tester.

    Who acts:SystemAIHuman

    Read more in the documentation

    Illustrative
    1. System Rules first environment healthy · locator changed Done
    2. AI Repair PR one line · locator value only Done
    3. Human Engineer approval repair PR waits for review Awaiting approval
    4. Human Bug draft real defects go to the tester to decide Awaiting approval

An illustrative walkthrough, not a recording

The screens above are drawn in HTML to show the flow; they are not captured from a customer system. For the details of each stage, read the documentation. To see the product live, book a 30-min call. On request we can open a time-limited, protected demo environment for your team.

Start with one story.

In a 30-minute call we look at your stack and a few of your stories, and tell you honestly whether a pilot fits.

Book a 30-min call Read the security overview