Zep

Overview

Zep is a long-term memory service designed specifically for AI assistants and agents. It provides conversation history management, entity extraction, and semantic search capabilities, making it easy to build AI applications that remember context over time.

Key Features

**Conversation Memory**: Stores and retrieves conversation history with automatic summarization

**Entity Extraction**: Automatically extracts and tracks entities from conversations

**Semantic Search**: Vector-based search over conversation history

**Memory Synthesis**: Generates insights from accumulated conversation data

**Fact Extraction**: Pulls key facts from conversations for quick lookup

When to Use Zep

Zep is ideal for:

AI assistants needing conversation history

Applications requiring entity tracking

Multi-turn conversational AI

Building agents that learn user preferences over time

Pros

Purpose-built for conversation memory

Strong entity extraction capabilities

Both open-source and hosted options

Good LangChain integration

Automatic conversation summarization

Cons

More focused on conversation than general memory

Self-hosting requires more setup

Some features only in cloud version

Getting Started

from zep_cloud.client import Zep

client = Zep(api_key="your-api-key")

# Add a conversation
client.memory.add(
    session_id="session-123",
    messages=[
        {"role": "user", "content": "My name is Alice"},
        {"role": "assistant", "content": "Nice to meet you, Alice!"}
    ]
)

# Search memory
results = client.memory.search(
    session_id="session-123",
    text="user's name"
)

Pricing

**Open Source**: Free, self-hosted

**Cloud Free**: Limited free tier

**Cloud Pro**: Usage-based pricing

**Enterprise**: Custom pricing

Memory Types