One memory.
For you and your agents.
Drop in. Remember.
A complete memory API
owned, verifiable, on Walrus
Capture anything in plain language,
notes and facts extracted into linked neurons automatically.
Store files on Walrus,
PDFs, docs, and folders chunked and indexed for recall.
Recall by meaning, not keywords,
two-stage retrieval with a cross-encoder reranker.
Grounded, cited, conflict-aware answers,
it surfaces contradictions instead of guessing.
Swap models, keep the context,
one owned memory — Claude, Gemini, and GPT stay in sync as you switch.
Plug into Claude, Cursor, any agent,
memory exposed as tools over MCP and A2A — model-agnostic.
Proactive insights while you're away,
sleep-time reflection synthesizes what matters.
Share memory across agents,
verifiable, access-controlled knowledge sets on Walrus.
Memory built for the agent era.
Owned, verifiable, persistent memory your agents can act on — fast retrieval, grounded reasoning, and proof it was never tampered with.
Beyond bi-encoders: cross-encoder reranking & contextual retrieval.
Why bi-encoders aren't enough, and how we pair a cross-encoder with Anthropic-style contextual chunking.
The architecture of Neurus.
Every memory is a neuron; links are synapses. How the layers fit — from ingest to proactive reflection.
Wiring Walrus, MemWal & your memory.
Bodies through MemWal, files on Walrus, a manifest you can publish and verify — and why the index is a cache.
Conflict-free shared memory with CRDTs.
An OR-Set op-log with Lamport clocks and capability signatures, so agents converge and bad edits are ignored.
The memory layer for the
teams building serious agents
Services