Enterprise agent architecture
Agents that read the enterprise's own semantic model, act inside enforceable boundaries, and leave evidence a reviewer can follow.
Enterprises possess enormous amounts of data, knowledge, process history and technical capability, yet struggle to assemble the right context, evidence and authority at the moment a consequential decision must be made.
Coming soonKindle
From systems of record to governed AI agents and enterprise outcomes
Most enterprise AI programmes stall at the same place. The demonstrations work. The pilot is convincing. Then someone asks what happens when the agent is wrong about a customer's contract, who authorised it to make that…
The enterprise does not make the model reliable. It makes the model's unreliability survivable: probabilistic reasoning sits inside deterministic control, governance and an evidence loop, and authority lives only outside the model.
You have to decide what the agent is allowed to touch, and defend that decision to people who will not accept "the model is usually right" as a control.
You own what runs in production. The interesting failures are not hallucinations; they are actions taken with authority nobody meant to grant.
You have to review a system whose behaviour varies by design. That is possible, but only if the evidence was designed in rather than reconstructed afterwards.
TOGAF, COBIT, ITIL, SAFe, programme management and enterprise risk each answer a real question about the enterprise. So do the AI management standards, and so do the vendor agent platforms. None of them answers the one that decides whether an agent should have acted.
LLM, RAG, agent, agentic system: every popular diagram stops at four rungs. The enterprise needs a fifth, where authority is checked before anything changes.
Letting a service desk assistant reset MFA is not phase two of the same programme. Once an agent can change state, the question becomes what it may change, and on whose authority.
I build, learn and write about complex technology systems — and, increasingly, about what it takes to make AI and agents genuinely useful inside an enterprise.
I’ve spent more than 20 years in software engineering, architecture, cloud, identity and large-scale transformation, across more than 100 enterprise projects. These days I build, test and write about agentic systems — and about how to make them genuinely useful in an enterprise while people keep control of authority, evidence and outcomes. Still learning. Still building. Still questioning assumptions.
A short note when the book is available, when companion material is released, and when a course opens. Nothing else.
Marketing consent is never required for companion-code access or support.