Why nexsus is a substrate

Every language model is brilliant and amnesiac — end the session and its world resets to zero. So we bolt memory on from the outside: a vector store here, a RAG pipeline there, scratch files everywhere. Glue, not foundation. Nexsus inverts that: memory isn't an add-on, it's the substrate the AI stands on — the AI substrate. For developers, why the word is earned: (1) Everything is one kind of thing — records, models, relationships, even the architecture itself are typed atoms in a single closed system; operations on the substrate return the substrate. (2) The substrate contains its own description — the model of the system is an object inside the system; the map of Nexsus is an element of Nexsus. (3) Library analogy — every record has a structure and a precise address, so you fetch the exact one by its code; and Nexsus embeds what each record means, so you can search by meaning ('unpaid invoices from hardware suppliers' with no matching keyword). One shelf: exact and semantic together.