Stratpoint Engineering

AI Foundational Training

Sign in with your Stratpoint Google account to continue.

AI SDLC Training
The Framework

The AI-Augmented SDLC

Same phases as always. AI generates, humans decide, and every artifact traces back to the one before it. This is the structure every module in the program teaches you to work within.

AI doesn't remove the phases of the software lifecycle — it removes the information loss between them. Each artifact is generated from the previous one and signed off by the human who owns that phase. The result is a single, traceable chain from business requirement to deployed code.

BRD→ PRD→ Architecture→ Dev Tasks→ Code→ Tests→ Deploy

The Workflow Diagram

The full pipeline — phases, roles, gates, and handoffs. The live diagram is embedded below; open it full-screen for walkthroughs during live delivery.

Open the diagram full-screen ↗

Same Phases, Reimagined

Across all six phases, the division of labor is the same: AI drafts from the upstream artifact, the named human owner reviews and decides.

1
Requirements
AI generates the PRD from an approved BRD  ·  Human SA / PM validate scope
2
Architecture
AI drafts the Architecture Document and ADRs  ·  Human SA owns every technical decision
3
Design
AI builds an interactive prototype  ·  Human UI/UX reviews implementation fit
4
Development
AI turns Dev Tasks into working code  ·  Human Tech Lead reviews alignment
5
Testing
AI generates test strategy and cases  ·  Human QA validates coverage and signs off
6
DevOps
AI configures CI/CD and production setup  ·  Human DevOps owns the release gate

What Holds It Together

Named human owners

Every artifact has exactly one owner who signs it off — not "the team." Accountability is the structure that makes the framework work.

Quality gates

Checked at every phase transition, not deferred to the end. A gate that fails sends work back to the current phase — not forward to the next.

Full traceability

Any line of code walks back to its BRD requirement. If it can't, the chain is broken and must be fixed before the gate passes.

Generator ≠ reviewer

The person (or agent) who produces an artifact is never the one who signs it off. Independent review is built into every handoff.

Four Non-Negotiable Principles

These govern every decision in the framework. No exceptions — not suggestions.

01
AI Augments, Humans Decide
Every AI output passes through human review before it moves downstream. AI never owns the decisions that carry formal accountability — the named artifact owner does.
02
Structure Before Speed
Invest in artifacts before engineering begins. Teams that skip upfront structure feel fast in Week 1 and stall in Week 3. The BRD and PRD are the substrate AI uses to produce useful architecture, tasks, and code.
03
Continuous Validation
Quality is checked at every phase transition, not accumulated and reviewed at the end. Every phase has an entry condition and an exit gate. Gate failed means back to the current phase, not forward.
04
Traceability as First-Class
BRD to PRD to Architecture to Dev Tasks to code to tests — every step references the one before it. A broken chain is a gap to fix before the gate passes, not a style issue.

The Universal Working Cycle

The same four-step cycle runs through every module and every task, at every scale. When something goes wrong in AI-assisted work, the diagnosis is always the same: which step was skipped?

01
Evaluate
Understand the task and its constraints before touching the AI. Read the Dev Task, the acceptance criteria, the ADR. Know what you're building.
02
Plan
Prompt for a plan first — not output. Review it against the spec. Correct misalignments at plan time, not after generation.
03
Apply
Execute, staying grounded in the plan. Redirect the AI if it drifts. You own every line — if you can't explain it, don't commit it.
04
Validate
Verify against the quality bar. Run the checklist, the tests, the gate. "Feels done" is not done.

One cycle, every scale

It's introduced in Module 2 and named in every module after. M2: artifact production. M3: prompt refinement. M4: session management. M5: implementation. M6: building assets. M7: knowledge contribution.