Thoughts on sovereign, sustainable AI.
Articles on AI architecture, RAG systems, data sovereignty and the practical implementation of AI in companies — pragmatic and without hype.
4 minCloudflare Clef: Decision Models for AI Workflows – with Self-Hosting
Cloudflare Clef and Clef-flash deliver structured decisions for AI workflows. Explore how they work and why running them on your own infrastructure matters.
Decision ModelsRead
6 minAleph Alpha Kolibri: German AI Model with Open Weights and Self-Hosting
Explore Aleph Alpha Kolibri: an open-weight AI model for German and English. Model comparisons, RAG, tool calling and requirements for self-hosting explained.
Open-WeightRead
12 minHermes Agent and alternatives: Which AI agent fits your work?
Compare Hermes Agent, OpenClaw, Bionic, Letta, Goose, Grok Bot, OpenAI Dots and Claude by task, memory, deployment model and control.
AI AgentsRead
13 minLocal Jev Alternatives: Which Small LLMs Work as a Fast Decision Layer?
13 local open-weight models tested as Jev alternatives: benchmarks, failed approaches and shadow data for a practical fast decision layer.
AI ArchitectureRead
11 minJev Explained: Why AI Applications Don't Need an LLM for Every Decision
What is Jev by TypeSafe AI? See how its typed decision model differs from LLMs, rules and classifiers, with practical agent and RAG use cases.
AI ArchitectureRead
7 minConnect a Document Management System to RAG
Connect a DMS to RAG with a robust architecture for parsing, OCR, versions, metadata, permissions, synchronization, hybrid search and citations.
RAGRead
12 minEnterprise AI Architecture Checklist: 12 Gates Before Build
Review enterprise AI architecture with 12 gates for business, data, identity, RAG, models, agents, APIs, security, evaluation and operations.
AI ArchitectureRead
6 minEnterprise RAG: Build, Buy or Hybrid?
Should you build or buy enterprise RAG? Compare data, permissions, retrieval, evaluation and operations with a layer-by-layer decision framework.
RAGRead
10 minAI Agent Evaluation: Execution Traces, Tool Calls & Production Readiness
Evaluate AI agents for production with execution traces, tool calls, guardrails, recovery, regression tests and cost and latency per task.
AI AgentsRead
6 minAI Agent vs Workflow vs Automation: Which Architecture Fits?
AI agent, LLM workflow or classic automation? Enterprise decision matrix for autonomy, risk, testability, tool calling and human-in-the-loop controls.
AI AgentsRead
7 minRAG Chunking Strategies for Enterprise Data
Choose RAG chunking deliberately: compare fixed, recursive, structural, semantic and contextual methods for PDFs, wikis, tables and DMS.
RAGRead
6 minHybrid Search vs Vector Search for RAG: When to Use Which
Hybrid search or vector search for RAG? Compare BM25, vector retrieval, RRF and reranking with a practical decision matrix for enterprise RAG.
RAGRead
8 minRAG Evaluation: Metrics, Golden Datasets & Regression Tests
Evaluate enterprise RAG: measure retrieval and answers separately, diagnose hallucinations and run golden dataset regression tests before go-live.
RAGRead
10 minPrivate AI vs public cloud: which enterprise AI architecture fits?
Private AI, public cloud, on-premise or hybrid? An architecture decision memo for workload placement, ownership, capacity, operations and exit.
Private AIRead
8 minPlan an AI pilot: from use case to a defensible go/no-go
How to plan an enterprise AI pilot: use case, data, scope, risk, architecture, test cases, success metrics, go/no-go and the path to production.
AI PilotRead
8 minRAG permissions: security trimming, ACLs & secure retrieval
How RAG permissions work: ACLs, user/group filters, security trimming, permission sync, revocation, tenant isolation and permission-aware evaluation.
RAGRead
9 minConnect SharePoint to RAG: architecture, Microsoft Graph & permissions
Connect SharePoint to RAG with Microsoft Graph, synchronization, parsing, versions, ACLs, security trimming, citations and production operations.
RAGRead
7 minConnect RAG data sources: SharePoint, DMS, ERP, CRM & databases
How to connect SharePoint, DMS, ERP, CRM, databases, file shares and APIs to a RAG system with synchronization, permissions, metadata, citations and evaluation.
RAGRead
5 minPrivacy Drift: Keep WordPress privacy changes visible
Privacy Drift shows technical privacy changes in WordPress: third-party resources, cookies, consent behaviour and deviations from a known baseline.
WordPressRead
8 minChatGPT & Copilot at Work: When Standard Products Are Enough
When do ChatGPT or Microsoft Copilot suffice at work—and when do you need a custom AI app? Decide by data, permissions, integrations and process logic.
EU AI ActRead
8 minAI Data Breach: What Really Counts in the First 72 Hours
AI data incident response: contain data flows, preserve evidence and check technical impact before privacy and legal teams decide next steps.
Incident ResponseRead
5 minThe EU AI Act for SMEs: What Really Applies to You in 2026
EU AI Act for SMEs in 2026: current duties, deadlines, AI literacy, transparency rules and prohibited practices – explained clearly.
EU AI ActRead
10 minPrevent Shadow AI at Work: Build an Official AI Path
Why shadow AI emerges and how companies can reduce private workarounds with clear boundaries, usable AI tools and an official AI working path.
AI GovernanceRead
8 minNot Everything Belongs in the AI Factory: Why Digital Sovereignty Is More Than Technology
Why digital sovereignty, local AI, and hybrid IT architectures matter more for European SMEs than blind dependence on hyperscale cloud.
Digital SovereigntyRead
9 minAI Sovereignty: Who Decides Which Models Companies Are Allowed to Use?
AI safety is necessary – but it must not become the private interpretive authority of a few providers. On access, control, and real agency.
AI SovereigntyRead
10 minGDPR-compliant AI for companies: what really matters
What does GDPR-compliant AI really mean — and where do your data actually end up? An honest guide for SMEs, without scaremongering.
GDPRRead
7 minWhy AI projects fail – and why the model is rarely the problem
Why AI projects fail: unclear business cases, poor data, missing evaluation, integration, security, operations and adoption – plus practical countermeasures.
AI strategyRead
13 minBuild a RAG system: architecture, retrieval, evaluation & operations
Build a production-ready RAG system: connect data sources, design retrieval, evaluation and permissions, and prepare citations and operations.
RAGRead