"The problem wasn't the AI. It was the plumbing."
The Core Thesis
In regulated AI, the moat is not the model — it's the governance layer. This insight, drawn from years of deploying AI systems in banking, insurance, public transport, and cultural heritage, became the foundation of the AI Plumber framework.
Most AI deployments start with model selection and end with a governance retrofit. In regulated environments — banking, healthcare, insurance, public sector — this sequence creates compliance debt that no organization can afford under the EU AI Act, SAMA, GDPR Article 9, or sectoral frameworks. The AI Plumber framework reverses this: governance becomes the first architectural layer, not an afterthought.
Production Experience
Before founding AI Plumber, Koen worked as AI Architect & Governance Advisor at Accenture Song Belgium, designing and delivering AI systems for regulated enterprises across Europe and the Middle East.
Najm Insurance — Saudi Arabia
Insurance Claims · SAMA ComplianceLed AI deployment processing 6,000+ daily claims under SAMA compliance and PDPL data protection across 40 cities. Zero-tolerance misclassification. Hybrid cloud + edge architecture with constrained agent identities per city deployment.
De Lijn — Belgium
Public Transport · EU AI ActDesigned the EU AI Act-aligned AI roadmap for Belgium's public transport operator. C-suite governance framework for 5,000+ FTE impact. 3-year roadmap approved at board level with 129% projected ROI.
US Restaurant Intelligence
Operational Intelligence · Multi-AgentArchitected a system that reduced a 200-person workflow to 3 AI agents. Processing time went from 1 month to 10 minutes. ~90% cost reduction with full audit observability and cost guardrails per agent.
NAMA Museum — India
Cultural Heritage · Sovereign DataBuilt sovereign data architecture for a €10M+ cultural heritage program. 180+ projectors, 99.9% SLA. Read-only + audit trail to preserve archival integrity and public accountability.
The Framework
Koen codified his production experience into four foundational patterns that form the backbone of the AI Plumber framework and the governance-first methodology used by enterprises moving from pilot to production:
Constrained Agent Identities
Every agent under a narrowly scoped service account. No inherited human privileges.
Attributable Actions
Every decision logged with full input context, reasoning trace, and output action.
Human-in-the-Loop Gates
High-stakes actions require explicit human approval before execution.
Kill Threshold Monitoring
Continuous telemetry. Threshold violations trigger automatic suspension + human escalation.
Writing & Speaking
"Governance-first is not a constraint on velocity. It is the only architecture that survives a regulator, a board, and a production incident simultaneously."