Portability / model-agnostic

What does model-agnostic mean for AI memory?

Model-agnostic memory is memory that is not tied to a particular model or vendor, so the same store works whichever AI you are using. Most AI memory is not portable: memory created inside one assistant lives inside that assistant. The consequence is lock-in by accumulation — the longer you use a tool, the more it costs to leave, not because the tool is better but because your context is trapped in it. Portable memory inverts that. The model becomes the replaceable part, and the accumulated context is the part you keep.