AI
AI engineering, written from production
Four subjects we work in daily, written down with positions. For technical decision makers and teams who want to know how this behaves in production.
Our AI knowledge comes from our own products, not a course: TenderScan.nl runs RAG with reranking in production, an AI-driven e-mail automation runs at a client through Supertaak, and our platform in development is driven primarily via MCP. What you read here, we have had to make work ourselves.
RAG implementation and rerankers
Why naive RAG falls short, what a reranker adds, and which trade-offs actually matter. With TenderScan.nl as production proof.
02Agentic workflows in production
What makes an AI agent reliable in production: restartable, scalable, auditable, with a human at the right moments.
03Integrating an LLM into your application
Model choice and switching, cost control, quality monitoring, and what you should not delegate to an LLM.
04MCP server development
Why the Model Context Protocol is becoming the integration layer between AI agents and existing systems, and what we expose with it.
Prefer the business perspective? That is on the service page LLM integration services.