Products·Implementation guide
Add governance to the agent you already built.
One session. No rewrite required.
Why this guide
You have an agent in production. It works. It does not have governance. You know it should - the scan told you that. This guide shows you how to add it without rewriting your system.
25 pages. Framework-specific sections for CrewAI, LangGraph, and Pydantic AI. Code examples you can copy. From scan results to working governance in one sitting.
What you get
The five properties
Identity, restraint, accountability, memory, and charter - implemented in order, with dependencies explained. Build them in the right sequence.
Framework-specific sections
CrewAI YAML configs. LangGraph state schemas. Pydantic AI tool definitions. Code that works in your stack, not generic advice.
Refusal implementation
Six refusal categories with real-world examples. How to build agents that know what they should not do - not just what they can.
Drift detection setup
Behavioral baselines, measurement cadence, and escalation thresholds. Know when your agent has changed before your users do.
EU AI Act mapping
Which governance properties map to which articles. The regulatory context for why this work matters beyond good engineering.
Verification step
Re-run the governance scan after implementation. See your score go from 1/5 to 5/5. Proof that the work landed.
Who this is for
Engineers with agents in production
Your agent works. This guide helps you make it governed. No rewrite required - governance layers onto existing architecture.
Teams adopting CrewAI, LangGraph, or Pydantic AI
Framework-specific guidance means you're not translating generic advice. The examples are in your stack's language.
Technical leads preparing for EU AI Act
Annex III obligations take effect 2 December 2027 under the May 2026 Digital Omnibus (revised from 2 August 2026; provisional political agreement pending Official Journal publication). Article 5 prohibitions and General-Purpose AI rules are live today. This guide maps governance properties to specific articles. Start now, not when the standards land.
Regulatory facts verified 2026-06-06.
Buy the guide
PDF download. 25 pages. Framework-specific implementation.
Get the GuideIncludes CrewAI, LangGraph, and Pydantic AI sections with code examples, the five-property implementation sequence, refusal categories, and EU AI Act mapping.
Go deeper
The guide gets governance into your codebase. If you need the complete practice, two doorways from here.
Trust Architecture Complete ($149)
All four trust architecture products - identity, voice, governance, and restraint - in the order they should be built.
View bundle →Governance Audit Engagement
We audit your agent system's governance architecture. Hands-on, in your codebase, with your team. Starting at $3,000.
Book discovery call →