BLOG
Notes from building AI platforms
Articles on what I learned building and deploying agentic platforms: analysis, diagrams, results and the code.
November 1, 2026 · 8 min read · Agentic AI · Platform · Governance
Building Enterprise Agentic Platforms: Lessons from NOVA
What changed when I stopped building an assistant and started building a team of agents that has to work inside a company.
November 1, 2026 · 8 min read · Operating model · Governance · Adoption
AI Operating Model: From Experiments to Enterprise Adoption
A practical path from isolated proofs of concept to shared platforms, with the governance that makes reuse possible.
November 1, 2026 · 8 min read · Evaluation · Multi-agent · Reliability
Evaluating Multi-Agent Systems in Production
Why a demo that works once says nothing about an agent system, and how I check that a new version is really better.
November 1, 2026 · 8 min read · Context engineering · RAG · Memory
Context Engineering: Beyond Traditional RAG
Retrieval finds text. Context engineering decides what an agent should know, from where, for whom and for how long.
November 1, 2026 · 9 min read · AI Factory · Operating model · Enablement
What Is an AI Factory? How to Set One Up Inside a Large Organisation
An AI Factory is not a lab and not a slide. It is an operating model that turns scattered AI experiments into products, deliveries and skills that repeat. Here is how I set one up at Devoteam, and what I would warn you about.
November 1, 2026 · 9 min read · Agentic AI · RAG · Decision guide
Agentic AI vs RAG vs Workflows: How to Choose
A practical guide to picking between a fixed workflow, a RAG assistant and an agent, based on what I saw building platforms for DILA and NOVA.
November 1, 2026 · 9 min read · Public sector · Sovereign AI · RAG
Sovereign Generative AI for the Public Sector: Lessons from DILA
What “sovereign” means for a public body deploying generative AI, and how a reusable platform, systematic evaluation and governance by design changed delivery at DILA.
November 1, 2026 · 9 min read · Governance · Agents · Security
AI Agent Governance: Policies, Human Approval and Signed Receipts
A chatbot that is wrong produces a bad sentence. An agent that is wrong produces a bad action. Here is how I think about controlling what agents may do, and proving what they did.
November 1, 2026 · 9 min read · MCP · Security · Context
MCP in the Enterprise: Connecting Agents to Confluence, Jira and Microsoft 365 Safely
The Model Context Protocol makes it easy to plug agents into company tools. Doing it safely is a question of permissions, classification and trust in retrieved text.
November 1, 2026 · 9 min read · Coding agents · SDLC · Governance
Earned Autonomy: Governing AI Coding Agents Across the SDLC
Instead of one global setting, review everything or trust the agent, grant autonomy per stage and per risk level, and only where automated checks can prove the work.
Figures come from validated client results and from my role at Devoteam. The projects on GitHub are public betas.