I built FortisNode after running privacy-focused infrastructure for 18 years.
Most AI providers promise privacy through their policies, but your data still reaches US servers, where it can be subpoenaed or accessed. FortisNode utilizes infrastructure-based privacy, featuring open-source models (Llama, Qwen, DeepSeek) that run entirely on European servers. Your data physically cannot leave the EU.
Technical details:
- Self-hosted LLMs on dedicated EU infrastructure
- No data retention, no training on user inputs
- Open-source models only (no Gemini/OpenAI/Anthropic API calls)
- GDPR-compliant by design, not by policy
Looking for feedback from the HN community on the approach and technical implementation.