The original title is "Owning your AI: why I self-host a personal assistant with long-term memory"

Short summary

The author argues for self-hosting a personal AI assistant to maintain data sovereignty — keeping memory, preferences, and conversation history in a local SQLite database on your own hardware. The assistant layer is decoupled from the underlying LLM so models can be swapped without data loss. The post references avelina.ai as an example but offers little architectural detail beyond high-level benefits.

• •Self-hosting preserves data sovereignty: memory stays in local SQLite, backups are user-controlled
• •Assistant layer is decoupled from LLM provider, enabling model swaps without data loss
• •Example setup runs in Telegram with calendar, notes, browser automation, and reminder skills

Generated with AI, which can make mistakes.