Mem0 Chat Memory icon

Mem0 Chat Memory

Persistent chat memory for AI Agent using Mem0 Self-Hosted OSS REST API

Mem0 Chat Memory

AI-generated

Summary

Provides a persistent chat memory backed by Mem0 self-hosted for use as AI Agent memory.

Inputs

  • userId (required) — Required memory owner user identifier
  • agentId (required) — Required memory owner agent identifier
  • runId — Optional conversation/session identifier
  • topK — Maximum number of relevant memories to retrieve per query (default: 10, min: 1, max: 50)
  • bufferLimit — How many latest user+assistant interactions to include in conversation mode (default: 20, min: 1, max: 200)
  • fallbackToSearchOnBufferMiss — When the buffer appears unrelated to the current query, fallback to Mem0 semantic search (default: true)
  • conversationRetrievalPolicy — How conversation mode chooses between buffer and semantic retrieval (options: smart_fallback, search_first, buffer_first; default: smart_fallback)
  • searchMode — Retrieval strategy used for search results post-processing (options: balanced, strict_facts, legacy; default: balanced)
  • maxContextChars — Maximum total characters injected into AI context after retrieval (default: 700, min: 100, max: 8000)
  • defaultQuery — Fallback search query when no user input is available (default: empty string)
  • rerank — Enable reranking in Mem0 search (default: false)
  • fields — Optional fields list for search response (comma‑separated string; default: empty)
  • includeAssistantMemories — Whether assistant messages should be included in retrieved context (default: false)
  • storeStrategy — How messages are persisted into memory (options: conversation, facts_only; default: conversation)
  • searchFilters — Optional JSON filters passed to Mem0 search payload (default: empty string)
  • allowEmptyContext — If disabled, retrieval automatically falls back before returning empty context (default: false)
  • infer — Legacy toggle kept for compatibility. Safe profile stores raw chat turns (default: false)

Output shape

AiMemory

The node returns a memory object that can be connected to an AI Agent's memory input. It does not produce any data on its own execute step.

Examples

Example 1: Use as memory for an AI Agent

Set User ID and Agent ID (required). Optionally set Run ID to isolate conversations. Adjust Top K, Buffer Limit, and retrieval policies as needed.

Links

Discussion