M.N.E.M.O. API
M.N.E.M.O. — short for Multidimensional Neural Elastic Memory Organization — gives AI agents exact, vector-free recall of what you store, and lets that same memory be published and subscribed to by other agents and teams. Dorothy is one thing built with it — not the only thing it can do.
Why M.N.E.M.O.
What you get
API access to the same memory and knowledge-sharing layer Panaxon AI builds on.Ingest documents and structured data, then recall or query it through a REST API designed for long-lived, multi-session agent memory.
In practice
Picture a travel agency's support agent recalling a customer's full booking history on a callback months later, instead of starting from scratch. That's one version of a pattern that shows up anywhere an agent needs to remember a specific customer, case, or record exactly — a support ticket, a client account, a claim. Memory built for a single conversation isn't enough for a real relationship. M.N.E.M.O. is the memory and knowledge-sharing layer for that pattern, not a finished vertical app. In practice, that looks like:
Any AI agent that needs to remember across sessions — not just within a single conversation window — can use M.N.E.M.O. as its persistent memory.
Because retrieval is indexed and exact rather than approximate, questions over ingested spreadsheets or records can return precise values, not paraphrased summaries.
A support tool could ground every conversation in a customer's full history — past interactions, preferences, prior issues — without asking them to repeat it.
Per-tenant isolation and role-based access rules make it possible to keep sensitive data scoped to the right roles as memory is shared across a team.
Share model
Exact retrieval is one half of M.N.E.M.O. The other half is letting that same memory be shared on purpose. You have knowledge — documents, notes, expertise — that people in your circle, or the public, would benefit from. Today that often means sending files, answering the same questions repeatedly, or hoping someone reads everything you wrote. With M.N.E.M.O., they just ask instead.
A person, team, or company can publish data to M.N.E.M.O. once. People who subscribe can query it through Dorothy chat, or through their own app connected via the M.N.E.M.O. API, like asking someone who already knows the answer.
Access is never automatic. Other M.N.E.M.O. users explicitly subscribe first, the same way you'd follow someone before seeing their updates. Circle-only data is shared with chosen people; public data can be subscribed to by anyone. In both cases, visibility starts only after explicit opt-in.
M.N.E.M.O. is the subscription and access layer, not the publisher and not the source-of-truth authority. The person or company publishing the data owns and is responsible for it. Subscribers know whose data they are querying.
This model supports both peer sharing (your circle) and public sharing (anyone who explicitly subscribes). The same publish flow serves both, with access rules set by the publisher.
Publishing flow
Publish knowledge from your Dorothy account settings — no API integration required.Beyond chat
M.N.E.M.O. isn't only a chat backend. MnemoRAG exposes the same exact, vector-free retrieval as a stateless RAG endpoint — one POST call to a named, fixed subset of your own subjects, with evidence from every subject in scope combined into a single synthesized answer. No chat session, no conversation history to manage.
For developers
Define a MnemoRAG endpoint in Dorothy Settings, then call it from your own pipeline.Proof of concept
Dorothy is Panaxon AI's own reference application — a conversational interface that shows what persistent, structured memory and knowledge sharing feel like day to day. It is one example of what M.N.E.M.O. can power, not a limit on what you can build with it.
Try it live
See M.N.E.M.O.'s memory in action in a live chat app.
Next step
Talk with Panaxon AI about API access to M.N.E.M.O.If you are building agents, copilots, or memory-heavy context pipelines, Panaxon AI can discuss fit, architecture, and access.
