I built an AI memory that says “not found” instead of guessing
Try it free: dorothy.panaxon.ai . Upload a document and ask it anything. For the past several months I've been building M.N.E.M.O. (Multidimensional Neural Elastic Memory Organization), a memory layer for AI that doesn't rely on vector search. The problem Similarity search always returns something , even when the answer isn't in your documents. The model then answers confidently from the wrong…
A new AI memory system called M.N.E.M.O. (Multidimensional Neural Elastic Memory Organization) has been developed, designed to avoid guessing when an answer is not found in the provided documents. Unlike existing systems that use similarity search, M.N.E.M.O. organizes data in a structured tree and retrieves information based on exact paths, not similarities.
Two key features of M.N.E.M.O. are its transparency and accuracy. Every answer generated by the system displays the exact path it was derived from, allowing users to verify the source. Additionally, when the system cannot find an answer, it explicitly states "not found" rather than providing an incorrect guess. This approach aims to prevent the AI from delivering wrong context in the first place.
However, M.N.E.M.O. is not without limitations. Its effectiveness relies on the underlying model choosing the correct path, and the system has not yet undergone large-scale benchmarking. Users are encouraged to try the system, report any errors, and share their experiences through a free account at dorothy.panaxon.ai. By doing so, the developers hope to identify and address any weaknesses in the memory organization.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
