AI Augmented Systems Architect

Architecture decision records an AI can resume from; zones drawn by responsibility and trust; seams with contract, direction and invariants; zero trust restated per seam, using bRRAIn's published eight-zone model as a worked example; and harvesting tribal knowledge into ADRs.

1. Pretest: ADRs, zones, seams and zero trust Diagnostic pretest · 5 min
2. The ADR anatomy — context, decision, alternatives, consequences Reading · 15 min
3. Zones as the unit of architectural reasoning Reading · 15 min
4. Seams — where zones meet and the contracts they trade in Reading · 15 min
5. Worked example — reading a published zoned zero-trust architecture Worked example · 15 min
6. Scenario — harvesting tribal knowledge before the agent re-litigates it Scenario · 15 min
7. Lab 1: From tribal knowledge to ADRs and a zone map AI role-play lab · 45 min
8. Retrieval: 8 questions across the module Retrieval check · 10 min

Mermaid, YAML and JSON Schema as architecture artefacts; the auto-derive pattern you build for your own system; one source per fact with CI consistency checks; layered context packs for agents; and a worked conversion from prose to a machine-readable set.

1. Pretest: machine-readable architecture Diagnostic pretest · 5 min
2. Mermaid diagrams as canonical architecture artefacts Reading · 15 min
3. JSON / YAML schemas for structural constraints Reading · 15 min
4. The auto-derive pattern — building architecture-now from your own code Reading · 15 min
5. One architecture, several formats — consistency checks and agent context packs Reading · 15 min
6. Worked example — from a prose architecture page to a machine-readable set Worked example · 15 min
7. Lab 2: One architecture in Mermaid, YAML and graph JSON AI role-play lab · 45 min
8. Retrieval: 8 questions across the module Retrieval check · 10 min

Functional and non-functional requirements as machine-readable specs; the 'AI says no' refusal pattern; performance, availability and security NFRs as testable invariants; zero-trust and audit invariants that cover AI agent callers; and evaluating constraint conflicts under delivery pressure.

1. Pretest: constraints, NFRs and zero-trust invariants Diagnostic pretest · 5 min
2. Functional vs non-functional requirements as machine-readable specs Reading · 15 min
3. The 'AI says no' pattern — agents refusing changes that violate constraints Reading · 15 min
4. Performance / availability / security NFRs as testable invariants Reading · 15 min
5. Zero-trust and audit invariants — making security architecture testable Reading · 15 min
6. Scenario — the agent says no, the deadline says yes Scenario · 15 min
7. Lab 3: Six NFRs as testable invariants with gates AI role-play lab · 45 min
8. Retrieval: 8 questions across the module Retrieval check · 10 min

Evolution-path notes that say what extends cleanly and what invalidates a decision; the four drift classes and how to detect each; the docs-first refactor protocol; architecture-now as a living document; and triaging a real drift report.

1. Pretest: evolution, drift and refactor Diagnostic pretest · 5 min
2. The evolution-path pattern — what extends cleanly, what doesn't Reading · 15 min
3. Drift detection — when documented architecture lies Reading · 15 min
4. The refactor protocol — update docs first, then code Reading · 15 min
5. The 'architecture-now' living document Reading · 15 min
6. Scenario — four drifts, one sweep Scenario · 15 min
7. Lab 4: Find the drift, write the refactor ADR AI role-play lab · 45 min
8. Retrieval: 8 questions across the module Retrieval check · 10 min

The pattern-match drill; trade-off matrices an AI can read and re-score; cost, risk, reversibility and novelty as triage axes that set AI-driven, pair-driven or human-driven ownership; and triaging competing requests into a decisive recommendation.

1. Pretest: pattern-matching, trade-offs and ownership Diagnostic pretest · 5 min
2. The pattern-match drill — does this fit, with what cost? Reading · 15 min
3. Trade-off matrix templates AI can read Reading · 15 min
4. Cost / risk / reversibility / novelty — the architectural triage axes Reading · 15 min
5. Scenario — three requests on a Monday morning Scenario · 15 min
6. Lab 5: Triage five inbound feature requests AI role-play lab · 45 min
7. Retrieval: 8 questions across the module Retrieval check · 10 min

AI-augmented design reviews where the agent presents and humans decide; recording and propagating decisions; living-set hygiene; storing ADRs in the bRRAIn vault with record_decision and write_vault_file and retrieving them with search_vault; and curating the knowledge graph with the Graph Editor.

1. Pretest: design review and architecture memory Diagnostic pretest · 5 min
2. The AI-augmented design review — agent presents the case Reading · 15 min
3. Humans decide; agent records and propagates Reading · 15 min
4. The living-set hygiene practice — weekly drift sweep, quarterly retrospective Reading · 15 min
5. Architecture memory — storing ADRs in the vault and retrieving them for agents Reading · 15 min
6. The knowledge graph as architecture-now for what your organization knows Reading · 15 min
7. Lab 6: Run an AI-augmented design review and record the decision AI role-play lab · 45 min
8. Retrieval: 8 questions across the module Retrieval check · 10 min

The capstone brief, how to size the architecture pack, the seven-dimension rubric and its hard fails, the four anchor exemplars, and a calibration lab in which you score a pack like the reviewer would.

1. Pretest: capstone preparation Diagnostic pretest · 5 min
2. The capstone brief — what's given, what's required Reading · 15 min
3. Capstone scope — sizing the pack so depth goes where it scores Reading · 15 min
4. Capstone grading — the seven-dimension rubric and how it scores Reading · 15 min
5. Anchor exemplars — what 92, 78, 71 and 54 look like Reading · 15 min
6. Lab 7: Score a capstone pack like the reviewer would AI role-play lab · 45 min
7. Retrieval: 7 questions across the module Retrieval check · 10 min

How the exam is built from the blueprint, pacing 59 items in 120 minutes, the most-missed item patterns, performance-task practice under exam conditions, and the capstone design review.

1. Pretest: exam readiness Diagnostic pretest · 5 min
2. How the exam is built — blueprint, item types and what each domain asks Reading · 15 min
3. Timing strategy — pacing 59 items in 120 minutes Reading · 15 min
4. The most-missed item types on architecture exams — patterns and recovery moves Reading · 15 min
5. Lab 8: Performance-task practice under exam conditions AI role-play lab · 45 min
6. Retrieval: 7 questions across the module Retrieval check · 10 min
7. Capstone: Architect and defend the Larkspur parts-and-claims subsystem AI role-play lab · 45 min