Memori
A LLM-agnostic agent memory layer for production systems, converting agent execution and conversation into structured, persistent states.
Who it is for: Developers who want self-hosting and control over the technology stack and data, and teams or organizations using production agents and requiring enterprise deployment forms
Core capabilities
- Structured Memory Capture: Captures each chat turn and classifies it as fact, preference, rule, or summary.
- Relevant Context Retrieval: Extracts relevant content from conversations and documents when context is needed in prompts.
- Semantic Context Augmentation: Automatically supplements semantic context for ambiguous language queries.
- Explainable Retrieval Results: Provides reasons for including retrieval results and traces relevance by entity, time, and source.
- SDK and MCP Access: The official website states that SDK can be integrated with one line of code into existing code; the pricing page lists Full SDK and MCP server access.
Pricing
Open Source is free for self-hosting; Cloud Free is a $0 hosted layer with a quota of creating 5000 memories and recalling 15000 memories; Team starts at $60,000 annually; Business starts at $150,000 annually; Enterprise is a custom quote.
Updated 2026-08-18