We adapt the TOGAF Architecture Development Method into a lightweight, structured flow that produces machine-readable artifacts AI agents can build from — and that governance forms fill from automatically.
The full Architecture Development Method — Preliminary plus phases A–H circling Requirements Management, across the Strategy, Core Layers and Implementation domains. AIAP keeps the rigor and distills the heavy lifting into six guided steps.
Set the org context, principles and the providers your agents may use.
Capture the headline requirement and target end state — the why before the how.
Name the actors, processes and outcomes the system has to serve.
Place the core services, agents and APIs that deliver each capability.
Declare where state lives and how it flows — the substrate everything runs on.
Pick the target hosting and guardrails; the result is a typed, governable architecture.
You answer plain questions about the system; AIAP turns the answers into typed components on the right layers.
Capture the headline requirement, target end state, stakeholders and goals — the why before the how.
Name the core services and where state lives; the canvas places them on the right layers automatically.
Pick the target hosting and guardrails; the result is a typed, governable architecture, not a slide.
Design discipline is the product: every architecture is versioned like software and assessed against the controls an auditor expects.
Publish v1, v2… with changelogs and restore any version into the working canvas — release discipline for architecture.
Solution Design, Architecture Review, DPIA and AI Impact fill from the graph and stay editable — mapped to GDPR, the EU AI Act and ISO 42001 controls.
Every change — by a person, the Copilot, or an MCP agent — is recorded with its source, so an auditor can trace exactly what happened and when.