← Elfege Systems / Field notesLet’s talk ↗

MOBIUS.ANAMNESIS

A memory of your own

Private AI memory that compounds.

Public brief · September 2026 · Product capabilities remain in development.

01 / The problem

A conversation ends. The work does not.

Decisions, reasons and lessons are scattered across sessions and documents. Starting every interaction without that history makes people repeat the same discovery work.

02 / The intended outcome

Bring relevant history into the next decision.

Anamnesis stores context on infrastructure you control and retrieves it by meaning. It is being built to support continuity across sessions, tools and people, while keeping the source of a recalled claim inspectable.

03 / The flow

Capture → retain → retrieve → use.

Eligible source material becomes a stored record. Retrieval selects potentially useful context for a question or session. The model uses that context to help answer; the original evidence remains important. Retrieval is not a guarantee of complete or correct recall.

04 / The privacy boundary

Storage and inference are different decisions.

Private storage can coexist with local or connected cloud models. A cloud-model integration sends the selected context to that provider. A fully local deployment needs local inference and compatible dependencies. Choose deliberately which sources are eligible for ingestion.

05 / The suite

Memory needs an operating environment.

The planned Core deployment brings memory together with coordination, session continuity and local access. Higher capability levels add agentic work and deliberation. Packaging and licensing are in development.

06 / Development status

Coming soon.

Anamnesis is used in internal development. Customer installation, reliability, documentation and support remain part of productization. We welcome investment, strategic partnerships and prospective pilot conversations; this is not a general-availability announcement.