Introduction

Memorer is production-grade memory infrastructure for conversational AI — sub-100ms semantic recall, automatic knowledge graphs, and zero-maintenance consolidation.

Why Memorer

Fast retrieval — Sub-100ms semantic recall, not keyword matching. Queries return ranked, relevant context ready to inject into your prompt
True understanding — Every remember() call extracts entities, relationships, and categories automatically. Your data becomes a structured knowledge graph, not a pile of embeddings
Production scale — Built to handle millions of users. Memory stays fast and accurate as your data grows
Zero maintenance — Automatic consolidation merges duplicates, resolves contradictions, and removes stale data. Your knowledge base stays clean without manual curation

How it works


from memorer import memorer
 
client = memorer(api_key="mem_sk_...")
user = client.for_user("user-123")
 
# Store a memory — returns IngestResponse with extraction counts
user.remember("The user prefers dark mode and uses VS Code")
 
# Recall relevant memories — returns QueryResponse with context string
results = user.recall("What editor does the user prefer?")
print(results.context)
# "The user prefers dark mode and uses VS Code"

The SDK is synchronous — no await needed.

Key concepts

Concept	Description
Memories	Individual pieces of stored knowledge (direct, derived, or inferred)
Entities	Named things extracted from memories (people, places, preferences)
Relationships	Connections between entities in the knowledge graph
Conversations	Sessions that combine short-term context with long-term memory
Consolidation	Automatic cleanup that merges duplicates and removes stale data

Next steps

Getting started — Install the SDK and store your first memory
Concepts — Understand the architecture
SDK Reference — Full API documentation
Integrations — Use with OpenAI, Anthropic, or LangChain