Memory that reasons.Answers that trace.
We are building the knowledge-graph memory layer for AI coding assistants: structured, traversable, and grounded in your real code.
Two failures, one cause.
Coding assistants kept failing the same two ways: they could not reason over how a codebase connects, and they could not show where an answer came from. Grep made the first worse and ignored the second. Vector search did the same thing with better manners.
Both follow from throwing the structure away before retrieval begins. So we built the thing that keeps it: a graph your assistant traverses instead of grepping, with every relationship carrying a tag saying how it was arrived at.
Three things, and we build to them.
Memory should be structured, current and on-device. In a world where assistants ship code, looks right is not good enough.
- Grounded over promises
- If we claim it, you can run the command and check it.
- Structure over similarity
- Meaning lives in the relationships, not in cosine distance.
- Open by default
- The core is open source. Trust is earned in the open or not at all.
Who is building it.
Graphify Labs, backed by Y Combinator, S26.