Remanentia

Remanentia is an evidence memory layer for local AI systems that need to remember operational context, retrieve sources, and verify answers before they leave the system. It is described as the memory layer of the Anulum agent runtime and is aimed at operators who want AI systems to use memory without turning internal records into public text.

The system turns scattered project knowledge into controlled evidence. It indexes files, documents, handovers, session logs, project facts, and selected corpora into a refreshable memory layer. For retrieval, it combines keyword search, embeddings, reranking, compiled facts, and direct recall paths to produce source-grounded answers. A factual guard verifies generated answers against retrieved evidence and blocks unsupported public output. It also has a public-safety boundary in which public APIs expose allowlisted corpora and redacted snippets while private operational memory remains internal by default. The page also says it ingests findings and decisions from the SYNAPSE coordination feed through an admission gate.

Remanentia is offered as a self-hosted memory service that can run beside existing agents and applications, with local-first inference and explicit hosted fallback modes. It also supports a controlled public API for approved corpora, verified snippets, and factual-control decisions. The page says it is part of the Anulum agent runtime, alongside Director-AI, Synapse, and HushLine, but Remanentia itself is the memory component. For commercial use, the verification layer referred to as Director Class AI is licensed for closed-source products or SaaS systems.