Personal AI memory
TwinAatma
Start the next conversation with some shared history.
A local memory for your AI tools: notes, decisions, and preferences you own, with proposed changes for you to review.
What should the next conversation remember?
A sample memory flow. Nothing here is saved to your device or an AI assistant.
When we compare options, start with a small working example.
A useful preference emerges during a conversation.
The problem
A new AI conversation often starts with explaining yourself again. TwinAatma gives an assistant relevant context from a collection of notes you can read and keep.
A design choice
Memory should be inspectable. Notes live in Markdown, and changes to the model of your preferences are proposed for approval. A lesson from one conversation can inform the next without silently rewriting who you are.
Where it stands
A public, local-first toolkit with an MCP memory interface and optional Obsidian or Notion workflows. It needs a compatible AI host and initial setup; this is a developer tool, not a hosted chat service.