The book

The model is the easy part. The plumbing is what kills you.

A governance-first field manual for shipping agentic AI into regulated production — the 6 Pipes, the implementation lifecycle, and four real regulated deployments, one per non-negotiable. No magic, just the plumbing the model doesn't give you.

In final consolidation · read the current manuscript free in your browser
Read free in your browser · or request a 4-page board briefing.
Front matter · two parts · a close

The manuscript, chapter by chapter.

Open with the failure modes, then the framework that fixes them — every pipe anchored to a real regulated deployment. Tap any chapter to read it in your browser.

Front matter
00
Introduction — Your AI Pipeline Has No Kill Switch

The model is the easy part. Pilots die at the governance layer, not the model layer — and the organizations that succeed aren't the ones with the best models, but the best plumbing.

Part One · The Problem
01
The AI Trust Crisis

Why enterprise AI fails — the real cost of failure, the three reasons it keeps happening, and the 2026 imperative now carrying €35M penalties.

02
The Experimental Trap

Chasing models instead of solving problems. The gap between lab and production, and how to break free of the trap.

03
What Regulated Industries Actually Need

The accuracy illusion, what "need" really means, the regulatory landscape, and compliance as a competitive moat.

Part Two · The Framework
04
Meet the AI Plumber · the 6 Pipes

The plumbing metaphor, and the six enforcement layers named: four non-negotiables, plus two that make a governed agent survive a real enterprise.

05
Governance-First Models · the 6 Pipes, enforced

Why ~70% of pilots die before production, ADRs as proof of rigor, the six pipes wired into the runtime, and the compliance mapping.

06
The Governance Playbook · the implementation sprint

A 30-day sprint from regulatory mapping to a production-ready, governed deployment — three phases and the sign-off checklist.

07
Case Studies in Governance Success

Four real regulated deployments — one per non-negotiable — plus the failure patterns and the Governance-First Manifesto.

08
The Future of Enterprise AI

Near-, medium- and long-term horizons — and the mantra that stays constant: infrastructure before intelligence.

Closing
Final Thoughts

Six pipes, four non-negotiable, each anchored to a real deployment. Build the kill switch, the ledger, the gate and the keyring first.

Useful before it’s finished. The moment a 4-page extract works, the book is already working.

Who it’s for

Compliance-minded operators

  • CDOs and CAOs deploying agents under banking, insurance, pharma or energy regulators
  • Architects who own the production system, not the pitch deck
  • Boards that need a governance posture they can put their name to

No hype, no “future of AI.” Evidence over adjectives.

Read it, then install it

The book and the layer

aiplumber.dev is the consultancy and the method. Kleiber is the open-source layer that enforces the four pillars in your agent runtime — the installable proof that the method holds.

Explore Kleiber & resources →

Regulatory currency: EU AI Act high-risk (Annex III) obligations apply from 2 August 2026 (operative date; subject to pending Digital Omnibus revisions — verify current status before relying on it).

Don’t wait for the last page.

Book a 20-minute call