Skip to content
mimagic
  • Services
  • How we work
  • Technology
  • About
  • Book a conversation
  • Deutsch

Insights

Insights.

What we think about AI, data and putting them to work. Including where we do not know the answer yet.


  • 5 May 2026 ·
    • sovereignty
    • eu-ai-act
    • architecture
    · 7 min read

    Sovereign Inference Is Not Sovereign Memory

    Mistral Medium 3.5 runs on 4 GPUs. That solves inference, not where your knowledge lives, who controls it, or how to prove it under the EU AI Act.

    Read post →
  • 4 May 2026 ·
    • research
    • world-models
    • agent-memory
    · 10 min read

    Beyond Retrieval: World Models for Knowledge Memory

    Retrieval finds stored items. Memory generalises from experience. The gap between them defines the next research problem in AI knowledge systems.

    Read post →
  • 1 May 2026 ·
    • rag
    • privacy
    • encryption
    · 7 min read

    Privacy in RAG Is an Architecture Problem

    PRAG proves E2E-encrypted RAG is technically possible. But it only protects embedding vectors, not your documents.

    Read post →
  • 1 May 2026 ·
    • agent-memory
    • rag
    • architecture
    · 7 min read

    Retrieval Is Not Memory

    Agent platforms call retrieval 'memory.' A new paper argues this is a category error, and the architecture choices get harder once you take it seriously.

    Read post →
  • 23 April 2026 ·
    • graphrag
    • explainability
    • auditability
    · 6 min read

    Graph Structure Is Not an Audit Trail

    A knowledge graph makes GraphRAG answers better, not auditable. XGRAG shows what auditability requires, and why the architecture sits outside the model.

    Read post →

Subscribe via RSS


mimagic
Book a conversation · Imprint Privacy

© 2026 MiMagic GmbH · Our own technology: Enchilada