Introducing Memoid: AI Memory Made Simple

Today, we’re excited to introduce Memoid — an open-source memory layer that gives your AI applications persistent, intelligent memory.

The Problem

Every AI developer faces the same challenge: AI applications are stateless. Each conversation starts fresh, with no memory of past interactions. Your chatbot can’t remember a user’s name from five minutes ago, let alone their preferences from last week.

Building memory from scratch is complex:

How do you extract relevant facts from conversations?
How do you store them efficiently?
How do you search semantically, not just by keywords?
How do you handle conflicting information?

The Solution

Memoid handles all of this with a simple API:

from memoid import MemoryClient

client = MemoryClient("your-api-key")

# Add memory from a conversation
client.add(
    messages=[
        {"role": "user", "content": "I'm a vegetarian who loves Italian food"},
        {"role": "assistant", "content": "Great taste! I'll remember that."}
    ],
    user_id="user_123"
)

# Later, search for relevant context
results = client.search(
    query="What food should I recommend?",
    user_id="user_123"
)
# Returns: "User is vegetarian", "User loves Italian food"

Key Features

Automatic Fact Extraction

Send any conversation to Memoid, and it automatically extracts the key facts. No need to parse or structure the data yourself.

Semantic Search

Search memories using natural language. Ask “What does the user like?” instead of searching for exact keywords.

Knowledge Graphs

Optionally extract entities and relationships. Know that “John works at Acme Corp” and “Acme Corp is a tech company” are connected.

Privacy First

Self-host on your infrastructure, or use our cloud with enterprise-grade security.

Get Started

We can’t wait to see what you build.

— The Memoid Team