Understanding local AI.

• What business AI really costs: subscription, API or own machine : Per-seat subscription, pay-per-token or owned hardware: public prices as of 1 October 2026, hidden costs, electricity and depreciation.
• Making an AI answer from your own documents: how RAG works : RAG explained step by step: chunking, embeddings, vector database, retrieval, citations. Its risks, and why sensitive documents are best kept local.
• Local AI: definition, how it works and concrete examples : What is local AI? Language model, weights, inference, hardware: how to run an LLM locally, what it allows and where its limits lie.
• Edge AI at work in the field, from the beehive to the workshop : Edge AI and embedded AI: latency, bandwidth, GDPR, sourced examples (VespAI, John Deere, implants) and a hybrid architecture for businesses.
• Hardware for local AI: sizing memory, bandwidth and power draw : Hardware for local AI: how much memory, what bandwidth, what power draw. Calculations, published measurements and examples, from Jetson to Mac Studio.
• Open source AI models in 2026: open-weight versus closed : Open source models in 2026: the measured gap with closed models, licences, DeepSeek, Qwen, Mistral, the hardware required and business uses.
• Local AI or cloud: where company data goes, and how to choose : Local AI or cloud: what happens to your data, what the Cloud Act, GDPR and AI Act require, and when to host the model in-house.
• GDPR and AI Act: checks before using generative AI at work : The AI Act after the July 2026 omnibus, CNIL recommendations, transfers, processors, logs, impact assessment: the points to check.
• Digital sovereignty and AI: when the cloud provider goes silent : Digital sovereignty and AI: AWS and CrowdStrike outages, Cloud Act, sanctions, AI Act. Dated facts and a method to reduce dependence on the cloud.