AIAP
PLATFORMMETHODMCPPRICINGABOUTSign in
AI-native enterprise architecturePart of the Inference Institute

Design architectures AI can build.

Governance paperwork is multiplying and AI made it worse — DPIAs, AI impact assessments and architecture reviews scattered across a dozen tools. AIAP collapses them into one living, machine-readable architecture that every assessment fills from, and that Claude and other agents read, extend and build over MCP.

Adapted TOGAFMachine-readableMCP-nativeVersioned
Get startedExplore the platform
aml-screening · v3live
Front DoorWeb · AppWorkersPostgreSQLAzure OpenAI
16 components17 connections4 layers
Part of the Inference Institute — architecting intelligence engineering success.
The problem

Governance is multiplying. Your architecture isn't keeping up.

Enterprises are drowning in design and governance artifacts — and AI just poured fuel on the fire. The documents that are supposed to prove your systems are safe and compliant are scattered across a dozen tools, drifting out of date the moment they're saved.

More forms than ever

Solution designs, architecture reviews, DPIAs, EU AI Act screenings, ISO 42001 and SOC 2 evidence — every initiative now ships with a stack of governance paperwork to fill in and keep current.

AI multiplied them

Each AI system drags its own compliance tail — model cards, AI Impact Assessments, risk reviews. The faster you ship AI, the faster the documentation backlog grows.

Scattered across tools

It all lives in slides, wikis, spreadsheets, tickets and drives — disconnected from the system it describes. Nobody can prove coverage, and every audit becomes an archaeology project.

AIAP collapses the sprawl into one living graph. Design the system once; every assessment and governance form fills from the same machine-readable source of truth — versioned, audited, and ready for an agent or an auditor.

DPIADriveAI Impact AssessmentConfluenceEU AI Act screeningSlidesISO 42001 controlsSheetsSOC 2 evidenceJiraSolution designWordArchitecture reviewEmailModel cardWiki
One architecture graphlive
B
D
A
T
versioned · governed · audit-ready
0+
Governance artifacts unified
DPIA, AI Impact, EU AI Act, ISO 42001, SOC 2…
0
Source of truth
One graph the docs fill from
0%
Changes audited
Person, Copilot or MCP agent
0
TOGAF layers
Business · Data · Application · Technology
Capabilities

An architecture that works as hard as your team.

AIAP makes enterprise architecture a first-class engineering artifact — structured, versioned, reusable, and consumable by AI. Not a picture of the system; the source of truth that builds it. Cards marked Soon are on the roadmap — including ideation, governed catalogues and gated reviews.

Structured canvas

Model systems as a typed node-graph — services, agents, models, datastores — mapped to architecture layers. The diagram renders the data; the data is the truth.

Versioned like software

Auto-save the working copy, then publish named versions with changelogs you can diff and restore. Enterprise architecture with release discipline.

Cross-project links

Reuse a component from another project, pinned to a version. When the source ships, dependents get an update prompt — provenance, never a silent break.

Library & templates

Start from curated reference architectures, drop in shared org components, and publish your own. Reuse with lineage instead of copy-paste.

MCP, read & write

Every architecture is a live Model Context Protocol endpoint. Agents read full context and propose changes through validated, audited tools.

AI copilot

An in-canvas assistant grounded in your project — it proposes components and connections as structured, reviewable diffs you approve.

Ideation workspaces

Soon

Spin up a free-form space to explore options and trade-offs with AI before committing anything to a governed canvas.

AI catalogue

Soon

A governed registry of approved models, providers and AI services — so every project builds on vetted, compliant building blocks.

Solutions catalogue

Soon

Curated reference architectures and solution patterns, ready to instantiate into a new project with lineage intact.

Data protection catalogue

Soon

Track personal-data flows, lawful bases and DPIAs in one place, linked to the components that actually process the data.

Change advisory

Soon

Route significant changes through a lightweight CAB with impact analysis and approvals — governance without the gridlock.

Configurable gated reviews

Soon

Require sign-off gates — architecture, security, DPIA — before a version can ship, configured per org and per risk level.

Method

An adapted TOGAF, built for the AI era.

We keep the rigor enterprise architects rely on — the four TOGAF domains, capability mapping, and governance — and shed the ceremony. Every component carries its layer, capabilities and metadata, so designs stay coherent as they scale and as agents extend them.

Capability tags, versioned snapshots and a full activity log give you TOGAF-grade traceability and reuse — best practices enforced by the tool, not a binder.

B
Business
Actors, processes and outcomes — the why behind the system.
D
Data
Stores, schemas and flows the system runs on.
A
Application
Services, agents and APIs that deliver capability.
T
Technology
Runtime, infrastructure and platform foundations.
Workflow

Design once. Implement with AI.

01

Design the architecture

Drag typed components onto the canvas, connect them with labeled edges, and capture goals, capabilities and decisions inline.

02

Version & reuse

Publish snapshots, link components across projects, and pull from the shared library so nothing is re-drawn from scratch.

03

Hand it to AI

Point Claude or any MCP client at the project. The agent reads the full graph and implements — no separate brief, no lost context.

Why it's different

A source of truth, not another document.

Diagramming tools and wikis give you pictures and prose that go stale. AIAP makes the architecture structured data — so governance, reuse and AI all work from the same place.

Slides, wikis & spreadsheetsAIAP
Machine-readable & parseable by AI
Versioned like software, with diffs
Governance forms fill from one source
Every change recorded for audit
Role-based access & encrypted secretsPartial
Agents read & safely extend it
Stays in sync with the real system
Model Context Protocol

Architectures your agents can build.

Generate a scoped token and any MCP client — Claude Code, Claude Desktop — connects to your project endpoint. Reads return the full architecture; writes go through validated tools. Every change is schema-checked, scoped to your membership, and recorded in the activity log.

Get your endpoint
claude · aiap mcpconnected
list_projectsread
get_architectureread
add_componentwrite
connect_componentswrite
update_componentwrite
→ validated · scoped · audited
Pricing

Start free. Scale when it ships.

Begin solo at no cost, grow into a per-seat team workspace, and add governance when you need it. Agents reading your architecture are always free.

Solo
Free

For individual architects exploring the platform.

  • 1 workspace, up to 3 projects
  • Full structured canvas
  • Community templates
  • MCP read-only access
Start free
TeamMost popular
$40/ seat / month

For teams designing and shipping AI systems together.

  • Unlimited projects & versions
  • Org library & cross-project links
  • MCP read + write, scoped tokens
  • AI copilot & provider connections
  • Activity log & audit trail
Start a team
Enterprise
Custom

For organizations with governance and scale needs.

  • SSO / SAML & SCIM
  • VPC / private deployment
  • Skills & custom MCP tools
  • Dedicated support & SLA
Talk to us

Indicative pricing for the v1 roadmap — final plans may change as features ship.

FAQ

Questions, answered.

Diagrams are pixels; AIAP architectures are structured data. Every node, edge, capability and decision is machine-readable, versioned, and exposed to AI over MCP — so the design gets implemented, not re-interpreted.
Assessments — Solution Design, Architecture Review, DPIA and AI Impact — read directly from the architecture graph instead of being re-typed into separate documents. One change to the system updates the source every form draws from, so coverage is provable and audits stop being archaeology.
The adapted TOGAF domains structure the design, and the built-in assessments line up with the controls teams report against — DPIA for GDPR, AI Impact / EU AI Act for AI risk, and the access, logging and data-protection evidence expected under ISO 27001 / 42001 and SOC 2.
Yes. Multi-tenant RBAC (owner / architect / viewer), row-level security on every table, AES-256-GCM-encrypted provider secrets, and a complete activity log that records every change — by a person, the Copilot or an MCP agent — with its source.
No. AIAP bakes an adapted, lightweight TOGAF into the canvas — four domains and capability tags — so you get the rigor by designing, without studying the framework or filling out templates.
Generate a scoped token and point any MCP client at your project endpoint. Read tools return full context; write tools add, connect and update components — each validated against a shared schema, scoped to your access, and logged.
AIAP is built by the Inference Institute — focused on the engineering practices that turn AI capability into shipped systems. The platform is how we architect intelligence engineering success in the open.

Architect intelligence that ships.

Design your system once, version it like software, and hand it to AI to build.

Get started freeSee pricing
AIAP
architecture.inference.institute
PlatformMethodMCPPricingSign in
© 2026 Inference Institute