The challenge
Make a large body of knowledge useful without sending documents to an external AI service. Keep sources traceable from the answer.
Dutch critical-infrastructure organisation
More than a million documents. One private search environment.
A fully disconnected RAG platform for searching and answering questions over more than a million documents. Designed to grow towards ten million.
What was built
GPU-accelerated document processing, hybrid search indexes, model serving and controlled access. The architecture connects retrieval, citations and quality checks.
The scale
More than a million documents in the environment. Ten million is the design horizon, not a claim about the current corpus. The ten layers on the technology page show our reference architecture.
VF / 01REFERENCE ARCHITECTURE
- People and applications: CHAT / API / DEVELOPER TOOLS
- Access · API gateway: IDENTITY / LIMITS / ROUTING
- AI agents, retrieval and models: TOOLS / SOURCES / MODELS
- GPU · Storage · MLOps: KUBERNETES / NVMe-oF / OBSERVABILITY
REFERENCE, NOT A NETWORK MAP
The pattern behind the solution.
A generic overview of the building blocks. Technical addresses, internal systems and client identities are excluded from publication.
Explore the architecture ↗