Tembory Memory icon

Tembory Memory

Advanced memory for AI agents with profile, tools, timeline, graph and semantic context

Tembory Memory

AI-generated

Summary

Retrieve and save advanced AI memory context using various retrieval modes, including semantic search, summaries, and hybrid methods, enriched with tool history, profile facts, decision and operational states, and more.

Inputs

  • authType (required) — Select authentication type for Tembory Cloud or Self-hosted instance.
  • threadId (required) — Stable identifier for the conversation or thread, used as the user ID if no explicit User ID is provided.
  • project — Stable namespace for the client or project to isolate memories between the same thread IDs.
  • retrievalMode (required) — Mode to retrieve context from memory; options include Basic, Summary, Semantic (v1 and v2), and Hybrid.
  • query — Natural language query used by semantic or hybrid search modes to find relevant memories.
  • memoryKey (required) — Key under which the memory context is returned.
  • payloadFormat (required) — Format of the context delivered to the AI agent before the LLM invocation.
  • advanced — Advanced options for fine-tuning retrieval, filtering, preset profiles, and returned data.

Output shape

a list with a single object containing the memoryKey with a list of system-role messages representing retrieved context from memory.

Returned data contains enriched context including vector memories, tool history, profile facts, decision and operational states, a textual summary, diagnostic data, and other memory-related metadata. Output is synchronized with connected AI models if present. Output is typically structured for AI agent consumption and conforms to the LangChain memory interface contract.

Links

Discussion