From the business department to the CIO, many feel the same pressure: „We have to do something with AI now." This whitepaper summarizes the key insights from the expert discussion between Prof. Dr. Ute Schmid (Universität Bamberg) and migRaven CEO Thomas Gomell on 9 December 2025 — a reality check for 2026.
The real problem: a flood of information instead of usable knowledge
Missing or incorrect AI answers are rarely an exotic special case. When knowledge is held chaotically, AI can only deliver good results to a limited extent: „If you ask bad questions, you get bad answers." The big data idea of simply storing everything takes its revenge — relevant content and data waste can hardly be told apart any more, and content production on autopilot aggravates the problem.
Humans and AI as a team
Where the standard workflow degenerates into „prompt, review, approve", professional skills erode (skill skipping). High performers, by contrast, use AI as a sparring partner — with human-in-the-loop structures that strengthen competencies instead of replacing them.
Knowledge graphs & neuro-symbolic AI
Dumping 20 years of documents into a vector database is not a strategy. Only curated knowledge models — knowledge graphs combined with language models — make AI explainable (answers can be traced back to nodes and edges), controllable (governance can be modeled) and sustainable (knowledge outlives a change of tools). A framework such as aikux.Brain serves as the foundation for solutions like migRaven.MAX.
Four fields of action
- Clean up and curate, 2. Anchor human-in-the-loop, 3. Pilot knowledge graph approaches, 4. Define sovereignty as a target metric — who controls our knowledge: we or our platform providers?
