Skip to main content
Story Creation & RefinementMany sources, one storyStakeholder review

Rough requirements in.
A testable story out.

Bring in Jira, Confluence, Figma, GitHub or a file, or just type the idea. The AI writes down what it understood, asks what is still open, and returns a user story with acceptance criteria and a quality score. Your team checks each step, then generates the test cases.

8
Sources in one story
A typed story plus Jira, Confluence, Figma, GitHub and files, summarised together
5
Clarifying questions
Spent first on the conflicts and gaps it found, with suggested answers to pick from
48
Hours a stakeholder link lasts
By default, with no account needed. Resend a new link if it runs out
How it works

Three steps, and you decide at each one

The same wizard whether you start from one typed sentence or several documents.

  1. Step 1

    Story Details

    Type the story or import it, give it a title and domain, pick a folder, and add more sources beside it.

  2. Step 2

    Questionnaire

    Read the draft story and the summary, fix what is wrong, then answer the questions or email them to a stakeholder.

  3. Step 3

    Refined Story

    Check the quality score, edit the story and its acceptance criteria, then go straight to test cases.

Step 1 · Story details

Start from wherever the requirement is written

Type the idea, or pull it in. The first import fills the form, and each one after that can be added as another source or replace the story. Up to 8 sources are read together.

  • Jira and Confluence

    Connect under Settings, then import an issue by its key or a page by its URL.

  • Figma

    Paste a file or frame URL with your access token, and the AI drafts a story from the design.

  • GitHub

    Give a branch or a pull request, and the story is written from what changed.

  • Files, sheets and screenshots

    PDF, Word, text, Markdown, PNG, JPG, Excel or CSV, up to 10 MB each. Drag one in and the AI pulls out the story.

Step 2 · Summary

One summary, every line tagged with its source

Before anything is refined, the AI writes down what it understood, so you check its reading first and not just the result.

  • Conflicts are listed, not settled

    When two sources disagree, both values and both sources are shown. The AI does not pick a side for you.

  • Gaps a tester would need

    Error handling, permissions, limits and empty states that no source specifies are marked for you to check.

  • Yours to correct

    Edit, remove or add lines, then resolve or dismiss each flag. Confirm summary records that you read this exact list, and any later change clears it.

  • Says what it could not read

    If a source is cut short or left out to fit one summary, you are told, so nothing goes missing quietly.

Step 2 · Questionnaire

Five questions, with answers ready to pick

A first draft of the story sits beside the summary. The questions then go to what is still open, so the answers settle something instead of repeating the requirement back.

  • A draft before any answer

    As a, I want to, So that, plus acceptance criteria, written only from your sources. Edit any line.

  • Questions go where they matter

    Conflicts and gaps first, then persona, success criteria, constraints, integrations and priority.

  • Pick, write or skip

    Choose a suggestion, write your own answer, add a note, or Skip All and refine with what you have.

  • Remembers similar stories

    What you settled on a similar past story can carry over as an assumption you can see and edit, instead of being asked again.

Stakeholder review

Get the answer in writing, from the person who knows

Not sure about an answer? Email the summary and questions from the Questionnaire step, or assign a stakeholder from History while the story has no test cases yet. Their answers stay on the story, not in a chat thread.

  • No account needed

    They open a secure link, answer the questions, mark each summary line Agree or Needs a change, edit the draft and add what is missing.

  • One link at a time

    A link lasts 48 hours by default. Resend emails a new one and the old link stops working, so one reply counts.

  • You review what comes back

    Check Stakeholder Response brings the reply in. Their edits and added lines wait for your review before the story is refined.

  • A late reply is not lost

    If they answer after test cases were generated, the story is flagged, and Regenerate with answers adds what they said.

Pending

Waiting for the stakeholder to reply.

Responded

The stakeholder has replied.

Expired

The link ran out first. Resend to ask again.

Late reply

They replied after test cases were generated.

Step 3 · Refined story

A story you can judge before you generate anything

The refined story comes back with its acceptance criteria, a quality score and an AI evaluation of what the answers clarified. When it draws on similar stories from your history, a badge says so.

  • Story Quality Score

    Clarity checks the As a, I want to and So that lines. Testability counts criteria with checkable wording. AC Coverage looks for seven criteria. Business Value reads the So that line. 80% or more reads as sprint-ready.

  • Estimated test cases

    A range worked out from the acceptance criteria, so you know what to expect before you generate.

  • Priority, with how sure it is

    Critical, High, Medium or Low, with a percentage for how sure the AI is. Hover to see what a priority is based on.

  • Edit before you generate

    Change the story lines, criteria and tags. Generate Test Cases takes the story as you left it.

In your team's control

It asks before it assumes

QAEverest does the reading and the first draft. Your QA team keeps the decisions, and at each point where a wrong assumption would carry forward, the page stops and asks.

  • Not a real story yet?

    Random keys, placeholder text or one word repeated are stopped before any AI runs. A thin story gets a warning, and you can still go on.

  • Title already used?

    A title that another story or test case record already has is caught, so each story keeps its own name.

  • Summary not confirmed?

    Refining asks first, so a summary nobody has read never steers the story silently.

  • Stakeholder hasn't replied?

    Generating test cases warns that their answers won't be in them, and offers Generate anyway or Wait.

After refinement

The story keeps working after you refine it

Generate the test cases

One click opens Generate Test Cases with the refined story and its criteria already filled in.

How generation works

Keep it in one folder

The folder you pick for the story carries over to its test cases, in the tree they share with your suites.

Trace it

Each refined story becomes a requirement row. Its criteria appear on the coverage report, and its priority sets the risk tier until a linked test has its own.

See traceability

Find it again

Refined stories save automatically. History lets you search them, filter by reply status, and reopen the files each one was built from.

More detail in the Story Creation FAQ and the case study on scattered requirements.

See It In Action

Type a raw story or idea below and preview the Story Details → Questionnaire → Refined Story flow.

Your refined user story will appear here.