Memory Chat
Ask about a concept in the loaded brain. Responses quote relevant stored experience context when available, otherwise they describe associations. These are not verified facts or LLM reasoning. If several concepts are mentioned, the longest matching name is used.
Send a prompt, edit memory using Learn, Experiences or Graph, then return and select Ask again / Compare. The original response stays fixed beside the latest response. Every request reads the currently active brain; chat never learns or saves memory.
History stays in this browser tab across navigation and reloads (up to 20 prompts). Closing the tab removes it. Each prompt is independent; previous messages are not conversational context.
Fuzzy search tolerates spelling mistakes when no exact concept matches. This setting is saved across refreshes; comparisons retain the original prompt's search mode.