Skip to main content

How I work

How I work

The process I apply to every requirement, the conditions a ticket must meet before it goes to the development team, and the principles I follow when AI takes part in the work.

The 6-step process

  1. Intake & clarificationMeeting minutes, Clarifying questions
  2. Analysis & designBRD / SRS, Wireframe, Business rules
  3. Handover to DevJira tickets, Acceptance criteria
  4. Testing & verificationTest cases, Bug reports
  5. UAT & acceptanceUAT checklist, Acceptance record
  6. Measurement & improvementWeekly report, KPI dashboard

Measurement & improvement → Intake & clarification · repeated steps become the standard for the next request

  1. Intake & clarification

    What I do
    Meet the client or read their request, then ask follow-up questions from a checklist: goal, users, business rules, exceptions, data, integrations. Record the meeting and add the transcript to the project knowledge vault.
    Output documents
    Meeting minutes, list of clarifying questions, confirmed original requirement
    Condition to move on
    The client confirms my understanding in writing
    Verifiable deliverable: Meeting Minutes & Question Log
  2. Analysis & design

    What I do
    As-Is/To-Be, impact assessment on existing modules by reading the codebase and the dev database, writing business rules, drawing user flows, wireframes and data mapping.
    Output documents
    BRD, PRD, SRS or mini-spec; user flow; wireframe; business rules table; ERD
    Condition to move on
    No open question left that affects scope
    Verifiable deliverable: BRD Excerpt & BPMN Flow
  3. Handover to Dev

    What I do
    Split Epic → Story → Task, write acceptance criteria in Given/When/Then form, attach flows and mockups, review the Definition of Ready with Dev.
    Output documents
    Jira tickets that meet the Definition of Ready
    Condition to move on
    Dev confirms they understand and can estimate
    Verifiable deliverable: User Story & Given/When/Then AC
  4. Testing & verification

    What I do
    Write test scenarios and test cases from the acceptance criteria. Agents run the cases that can be automated first; I verify failed and complex cases, reproduce and log bugs.
    Output documents
    Test cases, test report, bug reports with evidence
    Condition to move on
    No Blocker or Critical bug still open
    Verifiable deliverable: Test Execution & Bug Evidence
  5. UAT & acceptance

    What I do
    Prepare the UAT checklist, guide the client through UAT, turn feedback into tickets, close with sign-off.
    Output documents
    UAT checklist, acceptance record
    Condition to move on
    The client signs off in writing
    Verifiable deliverable: UAT Checklist & Sign-off
  6. Measurement & improvement

    What I do
    Track KPIs from Jira, hold a retrospective after the sprint, find repeated steps to standardise or hand to AI.
    Output documents
    Weekly report, KPI dashboard, process improvement proposals
    Condition to move on
    Back to step 1 with the next requirement
    Verifiable deliverable: Weekly Report & KPI Dashboard

Definition of Ready before handing over to Dev

A ticket is only pulled into a sprint when all nine items below are in place.

  • Business goal: why we are doing this, in one or two sentences
  • Users or roles affected
  • Business rules listed clearly, with codes (BR-01, BR-02…)
  • Main flow and exception flows
  • Wireframe or mockup if there is a user interface
  • Acceptance criteria in Given/When/Then form, testable
  • Impact on existing functions and data
  • Dependencies identified: APIs, data, third parties
  • Dev has read it and confirmed there are no remaining questions

Seven working principles

  1. Understand first, write later

    Do not start with a document. Start with the business problem and the real users.

  2. No vague requirements handed to Dev

    A requirement without enough context, business rules, expected behaviour and acceptance criteria does not go into the sprint.

  3. Every important requirement must be traceable

    From the business need to the design, tickets, test cases and acceptance.

  4. AI does the work, a person is accountable

    AI speeds up analysis and execution. Final responsibility lies with the person using the AI.

  5. Every AI output must be verified

    Requirements, mockups, SQL, test cases, diagrams and reports all need a person to verify them before handover.

  6. Automate repetitive work, not accountability

    Reasoning, decisions and accountability stay with people.

  7. Documents must serve delivery

    A document must help Dev build the right thing, Testers test the right thing and all parties understand the same problem.

Tools by step

Intake
Google MeetZoomrecording and transcriptsObsidian (project knowledge vault, wikilinks)
Analysis & design
Figmadraw.ioMermaidSQL on the dev databaseClaude Code to read the codebase
Handover
JiraPlaneTrelloMarkdown
Testing
Test case sheetsIDE agents running in parallel (Claude Code, Codex, Antigravity)Postman
UAT & reporting
UAT checklistJira REST APIKPI dashboardweekly report
AI for the team
BA KitTester Kit (agent skills)MCP (connecting Jira, databases and documents)

All case studies