HippocampAI

Enterprise Memory Engine for Intelligent AI Systems

HippocampAI is a production-ready, open-source memory engine that transforms how AI systems remember, reason, and learn from interactions. It provides persistent, intelligent memory capabilities with 120+ API methods covering knowledge graphs, hybrid retrieval, prospective memory, batch operations, multi-agent collaboration, and more.

Named after the hippocampus — the brain region responsible for memory formation — it gives AI systems human-like memory capabilities: storing preferences, tracking facts, detecting behavioral patterns, and delivering personalized experiences across sessions.

Current Release: v0.5.1

Prospective memory (time & event triggers), batch operations, deduplication API, RemoteBackend fixes, QueryRouter stem-prefix matching, and Groq timeout improvements.

Key Features

Knowledge Graph

Real-time entity & relationship extraction on every remember() call. Graph-aware retrieval with 3-way RRF fusion.

Hybrid Retrieval

Vector similarity + BM25 keyword + graph traversal combined via Reciprocal Rank Fusion for 40% better accuracy.

Multi-Agent Collaboration

Shared memory spaces for agent coordination. Multiple agents can remember and recall from collective knowledge.

Sleep Phase Consolidation

Bio-inspired memory consolidation: merge related memories, decay importance scores, prune low-value data.

Memory Triggers

Event-driven webhooks, websocket, and log actions that fire on memory creation, updates, or retrieval.

Procedural Memory

Self-optimizing prompts via learned behavioral rules. AI learns to communicate better over time.

Predictive Analytics

Pattern detection, habit tracking, and behavioral insights with cross-session temporal reasoning.

Auto-Healing

Automatic detection and repair of memory issues. Health monitoring and quality tracking built in.

Plugin System

Custom processors, scorers, retrievers, and filters. Custom schemas, tiered storage, export/import.

SaaS Platform

Full multi-tenant platform with auth, rate limiting, Celery background tasks, and React dashboard.

Prospective Memory

Schedule future actions with time-based or event-based triggers. Recurrence (daily/weekly/cron), priority ordering, full lifecycle management.

Batch Operations

Store/get/delete N memories in a single call. Deduplication with dry_run mode. Single-memory GET by ID.

Architecture Overview

HippocampAI follows a pipeline architecture with knowledge graph integration:

Ingestion
Embedding
Dedup
Knowledge Graph
Storage
Retrieval

Package Structure

HippocampAI is organized into two packages for flexibility:

  • hippocampai.core — Core memory engine (lightweight, no SaaS deps). For library integration and embedded use.
  • hippocampai.platform — SaaS platform (API, auth, Celery, monitoring). For self-hosted SaaS deployment.

Tech Stack

PythonFastAPIQdrantRedisOpenAIAnthropicGroqOllamaDockerCeleryReact

Why HippocampAI?

  • Open Source: Apache 2.0 license, free to use and modify
  • 120+ API Methods: Comprehensive coverage of all memory use cases
  • Production Ready: Redis caching (50-100x faster), 1000+ RPS, auto-healing
  • Multi-Provider: OpenAI, Anthropic, Groq, Ollama, or any custom provider
  • Privacy First: Self-hosted, your data never leaves your infrastructure