Building LLM agents, RAG, and knowledge graphs is my daily work. What I want from all of it is simple: to switch between models and tools freely without losing anything.
Memory as its own layer
That's why Eigen Mesh is a memory layer of its own, not a feature of one assistant. It connects your tools through MCP and gives you one place to keep your memories. It's LLM-agnostic and provider-independent, so every model and tool you connect reads from the same place.
Why owning it matters
I believe owning what accumulates in your head, and in your agents' heads, will become really important. So Eigen Mesh is open source, provider-independent, and fully yours.
Try it
The code is open source under Apache 2.0 at github.com/hikaru90/eigen, and the hosted version is at eigenmesh.xyz.