Hosted + Local Agent Memory
Persistent memory for AI agents, without inventing your own stack.
memnode gives agents long-term memory through a hosted API or a local MCP server. You get inspectable recall, provenance, scoped tokens, and shared tenants instead of stitching memory together from prompts and vectors by hand.
Use it for coding assistants, support agents, and research workflows that need durable recall, visible lineage, and a clean correction path across sessions.
When an agent recalls the wrong thing, you need the source, the correction chain, and the current winning memory. That is the real wedge, not just "persistent memory."
- Sign up with Supabase Auth and create an account.
- Provision one hosted tenant through the control plane.
- Create a scoped API token and call the Rust data plane over HTTP.
Typed provenance
Memories carry observed, reported, inferred, or hypothesized status instead of collapsing into one flat blob.
Hosted tenants
Provision a shared hosted tenant in the dashboard, mint API tokens, and keep quota and billing in one place.
Local MCP stays
The hosted product does not replace local usage. You can still run `memnode mcp` over stdio on your own machine.
Rust data plane
A signed-control-plane path, multi-tenant registry, and hot-path token cache keep hosted latency practical.
Example Shapes
Three concrete ways teams use memnode
Conversational assistant
Persist user preferences and past turns across restarts using the standard recall -> answer -> record loop.
Coding assistant
Store project conventions as typed entities instead of free-form notes, then recall them as structured context.
Research agent
Track claims with explicit provenance so later answers can separate reported facts from inferred conclusions.
Start free, keep local in reserve
The hosted SaaS is for fast onboarding, quota-managed tenants, and shared team workflows. The local mode stays available when you need offline or self-managed memory instead.
Latest writings
All articles →19 min read
How Memnode Evolved: From a Graph Database to a Memory Reasoning Engine
The honest engineering story of how memnode grew from a file-backed graph and embeddings store into a memory reasoning engine, reconstruc...
9 min read
Episodic and Semantic Memory: The Two-Layer Model Behind Durable Agent Recall
Why a flat vector index fails agents, and how memnode separates episodic memory (what happened) from semantic memory (distilled facts), l...
8 min read
Spreading Activation: Why Graph-Aware Recall Beats Top-K Similarity
Top-k vector similarity is the wrong default for agent recall: similarity is not relevance and it cannot follow the chain of facts a task...
9 min read
Canonization: How a Memory System Decides What It Believes
A naive store treats every recorded fact as equally true forever. memnode gives each memory a status that changes over its lifecycle: pro...
8 min read
Belief Networks for Agents: Holding Contradictions Instead of Overwriting Them
Most agent memory resolves conflicts by overwriting or silent top-k ranking, so an agent asserts a claim its own memory contradicts. memn...
8 min read
Sleep for Machines: Offline Consolidation in an Agent Memory Engine
Recall and writes happen on the hot path, but a memory engine needs a slow clock too. How offline consolidation promotes episodes to fact...