Hi, I'm Nikolaj Fløjgaard.
Senior Integration Architect writing about AI agents, systems, autonomy, and the infrastructure behind what actually works.
Less hype, more systems thinking: practical notes on enterprise integration, large-scale transformation, agentic workflows, autonomy, and the infrastructure behind what actually works.
Start from the right path
Different people land here for different reasons. This should make that easier.
Enterprise architecture and transformation
For people dealing with architecture, platform decisions, integration mess, and large-scale transformation pressure.
See work
AI systems and agentic workflows
For people trying to separate real operational AI from prompt theatre and demo energy.
Explore AI writing
Built systems and proof-of-work
For people who want to see actual artifacts, experiments, integrations, and product-shaped systems.
See projects
Recent writing
Latest posts, if you want the freshest signal instead of the curated entry points above.
The point of an MCP platform is not more code
AI can generate more code. That is not the bottleneck. The real work is removing integration, ownership, approval, and governance bottlenecks with a domain-driven MCP platform, APIM, and MLflow from mlflow.org.
I made my agent's memory faster by making it smaller
The useful version of agent memory is not a bigger bucket. It is a layered system: compact truth, raw notes, topic summaries, and a local index that works even when embeddings do not.
Daniel Kokotajlo is probably right, just too early
Daniel Kokotajlo's AI forecasts look extreme until you separate the mechanism from the calendar. The direction is plausible. The 2027-style timeline is the part I struggle to buy.