Knowledge

AI and knowledge management: fundamentals, architecture, practice

How companies make knowledge findable, keep it secure and put it to work with language models: from retrieval-augmented generation through semantic search and document AI to permissions and checking AI answers. Each page opens with a short direct answer and lists its sources.

Knowledge management Knowledge management with AI: fundamentals, architecture and practice Knowledge management with AI explained: explicit and tacit knowledge, the lifecycle, AI knowledge base architecture, ISO 30401, maturity and limits.
Retrieval-augmented generation Retrieval-augmented generation (RAG) explained: pipeline, variants, limits How RAG works: the pipeline from ingestion to citation, chunking strategies, hybrid search, reranking, typical failure modes and variants.
Semantic search Semantic search in the enterprise: full text, vectors, hybrid Semantic search explained: BM25 versus vector search, hybrid search with Reciprocal Rank Fusion, German compounds, and measuring relevance with Recall@k.
Document analysis How AI reads documents: OCR, layout and visual search How AI reads documents: the classic OCR pipeline, layout-aware models, visual search like ColPali, tables, scans and how to measure quality.
Knowledge graph Knowledge graphs and GraphRAG: when relationships matter more than similarity Knowledge graphs and GraphRAG explained: RDF versus property graphs, ontologies, extraction, when relationships beat vector search and what upkeep means.
Open-weight model Local language models: what open-weight models can do and what they need Local language models explained: open weights versus open source, quantisation, memory sizing as a worked calculation, serving, operations and limits.
AI agent AI agents in knowledge work: from retrieval to action AI agents in knowledge work explained: the agent loop, tools, the Model Context Protocol, agentic retrieval, approval points and failure modes.
Knowledge transfer Capturing knowledge before it retires: knowledge transfer with AI How mid-sized companies capture experience-based knowledge before retirement: methods, a handover checklist, roles, and what AI can and cannot do.
Hallucination (artificial intelligence) Checking AI answers: hallucinations, source grounding and evaluation Checking AI answers: error types, source grounding with citation checks, test questions, metrics such as faithfulness, human spot checks and monitoring.
Access control Access control in AI knowledge systems: who gets to see which answer How access rights travel through a RAG pipeline: carrying document permissions, filtering at query time, connecting identity, data classes and logging.

For readers moving from understanding to building: software development and the series From Prototype to Production. For the question of whether and where AI makes sense for your company: KI-Beratung.

Let us talk about your project

Whether it is a prototype, an internal tool or an AI application: describe briefly what you are building or want to take into production.

Get in touch See AI development Or: request a 20-minute first call