Ori Mnemos FAQs

Question 1

What is Ori Mnemos and how does it benefit AI agents?

Accepted Answer

Ori Mnemos is a persistent, markdown-native memory infrastructure that enables AI agents to maintain their identity, knowledge, and operational state across sessions. This prevents agents from starting from zero context, ensuring continuity and more effective long-term operation.

Question 2

What makes Ori Mnemos's information retrieval unique?

Accepted Answer

Ori Mnemos uses a 'Three-Signal Retrieval Engine' that fuses semantic embeddings, keyword (BM25), and graph (PageRank) signals. This combines different aspects of relevance to surface highly pertinent information, adapting dynamically based on query intent.

Question 3

Does Ori Mnemos require a database or cloud service?

Accepted Answer

No, Ori Mnemos is entirely markdown-native, storing memory as plain files on disk. It builds an in-memory knowledge graph and uses local SQLite for embeddings, eliminating any database or cloud dependencies and ensuring data ownership and privacy.

Question 4

How does Ori Mnemos manage the decay of knowledge over time?

Accepted Answer

It incorporates a 'Vitality Model' that explicitly tracks note decay based on access frequency, structural connectivity, and metabolic rates. This model can identify 'fading' knowledge, prompting agents to review, reconnect, or update information to maintain its relevance.

Question 5

How does Ori Mnemos ensure an AI agent's identity and continuity?

Accepted Answer

It features an Agent Identity Layer that automatically injects an agent's identity, goals, and methodology at connection. Agents can also update their own identity files, ensuring consistent self-awareness across sessions and deployments without manual configuration.

Ori Mnemos

Ori Mnemos

Key Features

Use Cases

Key Features

Use Cases