Agentic FX Dev Squad · Training Programme

Domain Owner
Transition Training

Structured curriculum for Manual QC, Automation QC, and Leader roles transitioning into the Domain Owner position. Five modules covering mindset, documentation, spec governance, validation, and roadmap ownership.

Target audiences
Manual QC
Automation QC
Leader
Training modules
Module 01 · Foundation
The Domain Owner mindset
  • What "owns the business layer" means in daily practice
  • Shift from verifying output → defining expected outcomes upfront
  • Why domain knowledge must be documented, not just held in memory
  • Your authority: scope decisions, AC correctness, business logic disputes
  • Where Domain Owner ends and AI Engineer begins — the decision boundary
Module 02 · Operations
Documentation, governance, validation & strategy
  • What goes in business-rules, glossary, and personas files — and when to update them
  • How AI agents read and use your domain files to generate specs
  • How to read an AI-generated spec.md with a critical business lens
  • Reviewing acceptance criteria for correctness — not technical soundness
  • When to approve, send back, or escalate; what status: approved commits you to
  • What "manually walk each P0 AC" means — not running scripts, not spot-checking
  • Distinguishing a business logic failure from a technical defect
  • Contributing priority ordering and urgency to the roadmap (not effort estimates)
  • The ship gate: what "all P0 ACs ✓" sign-off actually commits you to
  • Scope vs. out-of-scope: documenting decisions and their rationale
Case Study · Applied practice
JEC-3620 — ABCA Contract Filter
  • End-to-end walkthrough of a real Domain Owner ticket from intake to ship
  • See how business rules, glossary, and personas feed an AI-generated spec
  • Review acceptance criteria with a business lens — what to approve vs. send back
  • Apply concepts from Module 01 (mindset) and Module 02 (operations) in context
Decision boundaries the Domain Owner owns
Decision Owner Notes for trainees
Is this idea worth speccing?
Both
DO on business value; AI Engineer on feasibility. You hold the value half.
What's in scope vs. out of scope?
Domain Owner
Sole decision — AI Engineer flags implications but does not decide.
Is the spec ready to approve?
Both
DO gates on business correctness; AI Engineer gates on technical clarity.
Which AC was misunderstood?
Domain Owner
You are the sole authority on intent. Resolve in Phase 4 review findings.
Is the code architecturally sound?
AI Engineer
This is explicitly not your call — build trust by not overstepping here.
Do tests cover enough business scenarios?
Both
AI Engineer on coverage percentage; DO on whether the scenarios reflect reality.
Is the feature ready to ship?
Both must sign off
Your gate: all P0 ACs checked. AI Engineer gate: all tests green.
Audience transition paths
MQC

Manual QC → Domain Owner

Closest background · highest AC intuition
Unlearn
Your role is no longer to catch bugs after the fact — it is to define what "correct" means before a line of code is written.
Leverage
Deep AC intuition transfers directly. Use it to write sharper criteria and immediately spot vague or untestable specs.
Extend
Learn to capture business rules as structured domain files, not just test case comments or tribal knowledge.
Practice
Shadow a full spec review cycle end-to-end before flying solo on the approval sign-off.
AQC

Automation QC → Domain Owner

Strong structure · needs scenario reframing
Unlearn
Your value is no longer in the scripts. Agents generate test code. Your unique contribution is scenario completeness judgment.
Leverage
Structured coverage thinking transfers to reviewing whether test-case-designer covered all real business scenarios.
Extend
Shift from "does this test pass?" to "does this scenario represent a real user situation with real consequences?"
Practice
Review an AI-generated test plan and flag missing business edge cases. Repeat this drill across at least three features.
LDR

Leader → Domain Owner

Strong stakeholder fluency · needs detail-level discipline
Unlearn
Coordination and status-reporting are increasingly handled by agents. Your unique contribution is business judgment and priority framing — not orchestration.
Leverage
Stakeholder communication skills translate directly into writing clear backlog problem statements and reviewing release summaries for accuracy.
Extend
Move from high-level direction to ground-level spec correctness. You need to read and approve at detail level, not just endorse at summary level.
Practice
Own the scope/out-of-scope decision for a real backlog item — then document the reasoning in docs/domain/ so agents can use it.
Readiness checklist — exit criteria
Can write a complete domain doc — business rules, glossary, personas — without prompting or referencing examples.
Can review a spec.md and identify at least 3 types of business logic errors before granting approval.
Can distinguish which review findings are "business logic / AC misread" versus technical issues — and route accordingly.
Has walked P0 ACs manually on at least one complete feature and checked off acceptance-criteria.md independently.
Can articulate scope decisions — what's in and out — with documented business rationale in domain files.
Understands ship gate responsibility — approving "all P0 ACs ✓" is a binding sign-off, not a rubber stamp.
⚡ AI agents automate spec drafting, test generation, code writing, and code review — but they do not make judgment calls about business value, scope correctness, or AC intent. Every module in this curriculum builds the judgment that AI cannot replace.