Nomi’s Mind Map made part of companion memory inspectable

A visible summary layer

Nomi’s October 9, 2025 Mind Map 2.0 announcement describes an interactive graph of people, places, topics, and goals. The company says it draws from long-term memories to form high-level overviews, while individual memories remain separate. Users can inspect evolving nodes and links, and can add entries; according to the announcement, the first map may take roughly 500 or more messages to form. This is an implementation description, not a claim that every remembered detail is surfaced.

The March memory update gives the backdrop: Nomi had already described expanded medium-term retention and long-running narrative continuity. Mind Map changes the presentation and control question. Rather than only asking whether a companion recalls something in a reply, a user can see a company-selected abstraction of what the system regards as related. That makes memory legibility a product surface in its own right.

What visibility does and does not mean

Evidence: Nomi says maps coexist with rather than replace memories, and that they update as conversation continues. Inference: a visual map may make a mistaken association easier to notice and discuss, but it should not be mistaken for a complete data export or an explanation of every model decision. The post does not establish whether every influencing memory appears, how long all records persist, or whether edits propagate in a predictable way across future responses.

Evidence limit: The feature’s coverage and operational effects are described only by Nomi; no account-level inspection occurred.