Neotoma vs platform memory
Platform memory is the built-in memory layer inside products like Claude, ChatGPT, Gemini, and Copilot. It is optimized for convenience inside one vendor surface. Neotoma is a deterministic state layer optimized for portability, auditability, and exact state reconstruction across tools.
Both platform memory and Neotoma help agents remember across sessions. They target very different reliability and control requirements.
How does Neotoma compare to platform memory?
Platform memory
Platform memory is controlled by the model provider. It may remember preferences, summaries, or user details inside that product, but the storage model is opaque. Users typically cannot inspect lineage, replay state transitions, or export a deterministic observation log.
Neotoma
Neotoma stores append-only structured observations with deterministic reducers, schema validation, and provenance. Memory remains user-controlled, cross-tool, and reconstructable at any point in time.
Guarantee comparison
When to use which
Use Platform memory when
You want the fastest possible setup inside a single AI product, and convenience matters more than portability, versioning, or auditability.
Use Neotoma when
You need memory that survives tool changes, supports exact historical reconstruction, exposes provenance, and gives multiple agents a shared state layer with formal guarantees.
Common questions
Can I use platform memory and Neotoma together?
Yes. Platform memory can hold lightweight in-product preferences or convenience context, while Neotoma stores durable structured state that must persist across tools and sessions.
Does platform memory provide auditable history?
Not in the way production agent systems usually require. Platform products may expose some editable memory UI, but they generally do not provide append-only observation logs, deterministic replay, or field-level provenance.