The Private AI Assistant With No Cloud Dependency — Complete and Working

Docker Model Runner's network activity is limited to your local machine by default — you can verify this with network monitoring tools. OpenClaw routes messages through your local model endpoint rather than external APIs. The instructor covers how to verify the no-cloud-dependency of your setup during the workshop using practical monitoring techniques.

None for AI inference. Docker Model Runner runs the LLM entirely locally. OpenClaw processes messages entirely locally. The only external network usage is for WhatsApp and Telegram message delivery — which is the same as any messaging app you use — and initial model downloads from Docker Hub.

Yes. Many developers use this setup in corporate environments specifically because of the no-cloud-dependency guarantee. Your conversations never leave your corporate network. The instructor covers security configuration best practices for corporate deployment during the workshop.

Model downloads happen from Docker Hub which is external. Once downloaded, the model runs entirely locally with no external calls. You can choose to disable automatic updates and update models manually if you need to guarantee complete network isolation during operation.

Performance depends on your hardware. On a laptop with 16GB RAM and a 7B parameter model, expect 10 to 25 tokens per second on CPU — perfectly usable for personal assistant conversations. The instructor covers performance expectations and hardware-appropriate model selection during the workshop.

Yes. Both OpenClaw and the open weight models used in this workshop are open source. You can inspect the code to verify what network calls are made and confirm the no-cloud-dependency of the system. This level of auditability is not possible with proprietary cloud AI services.