Guide
Python SDK
JS SDK
HTTP API
MCP Server
Guide
Python SDK
JS SDK
HTTP API
MCP Server

Introduction to Memof.ai

Because your AI deserves a memory better than a goldfish 🐠➡️🐘

Welcome to Memof.ai (Memory of Agents), where we solve one of AI's most frustrating problems: remembering stuff. You know how ChatGPT forgets your conversation the moment you close the tab? Yeah, we fixed that.

What is Memof.ai?

Memof.ai is a persistent memory platform for AI agents. Think of it as a brain transplant for your AI—but without the sketchy surgery.

Regular AI: "Who are you again?" 😕 AI with Memof.ai: "Oh hey! You're Sarah, you prefer dark mode, and you're working on that React project!" 😎

The Problem We Solve

Building AI applications is awesome until you realize:

  • 💔 Conversations reset every session
  • 🤷 Context gets lost between interactions
  • 😤 Users have to repeat themselves constantly
  • 🔥 Personalization is impossible without memory

We've all been there. You build a cool AI chatbot, users love it, then they close the browser and... poof... everything's gone. Like it never happened.

Not anymore.

How It Works

The simplest explanation you'll ever read 📚

The Magic Formula

Your AI + Memof.ai = AI That Actually Remembers Things™

Here's the flow:

  1. User interacts with your AI agent
  2. Store important bits in Memof.ai (preferences, facts, context)
  3. Retrieve memories when user comes back
  4. Profit (probably—we're not financial advisors)

The Architecture

┌─────────────┐ │ Your App │ "Store this!" └──────┬──────┘ │ ▼ ┌─────────────┐ │ Memof.ai │ 🧠 Semantic Memory Storage └──────┬──────┘ │ ▼ ┌─────────────┐ │ AI Agent │ "I remember you!" └─────────────┘

It's that simple. No PhD required.

What Makes Us Special?

Glad you asked! 🌟

1. Semantic Search

We don't just store text—we understand meaning. Search for "coding preferences" and find "likes Python" or "prefers functional programming."

# Store this
"User prefers dark mode and writes Python"

# Find with this
client.memories.search(query="programming language preference")
# ✅ Found! Even though you said "Python" not "programming language"

2. Multi-Language SDKs

Python? JavaScript? HTTP? We speak your language.

  • 🐍 Python SDK - Type hints included, mypy approved
  • 🟨 JavaScript/TypeScript SDK - Full type definitions
  • 🔗 MCP Server - For Claude, GPT, and friends
  • 🌐 HTTP API - For that one project in Go/Rust/PHP/whatever

3. Organized by Design

Workspace → Bots → Memories └─ Your Project └─ Customer Support Bot └─ "John prefers email" └─ "Sarah likes dark mode"

Everything has its place. Marie Kondo would be proud.

4. Battle-Tested

Used in production by companies who can't afford memory loss. (We won't name names, but they're impressive.)

Real-World Use Cases

Because "use cases" sound professional 💼

Customer Support

Problem: Agents forget customer preferences
Solution: Store and retrieve customer context
Result: Support that feels personal, not robotic

// Before
Agent: "Hi, how can I help?" (for the 47th time)

// After  
Agent: "Hi Sarah! Following up on your API integration issue..."

Personal AI Assistants

Problem: Your AI has amnesia
Solution: Persistent memory across sessions
Result: An assistant that actually knows you

# Remembers your preferences
"You usually prefer TypeScript over JavaScript"

# Recalls your context
"Last time you were working on the authentication module"

# Tracks your projects
"Your deadline for the API docs is next Friday"

Chatbots

Problem: Users hate repeating themselves
Solution: Store conversation context
Result: Conversations that flow naturally

User: "I'm interested in your API"
Bot: [stores: interested_in = API]

// Two weeks later...
User: "Hey, I'm back"
Bot: "Welcome back! Ready to dive into that API integration?"

AI Agents

Problem: Agents can't maintain state
Solution: Persistent agent memory
Result: Smarter, more capable agents

Who's This For?

Hopefully you! 🎯

  • 🧑‍💻 Developers building AI applications
  • 🏢 Companies needing persistent AI memory
  • 🚀 Startups building the next big thing
  • 🤖 AI enthusiasts who love cool tech
  • 👨‍🔬 Researchers experimenting with agent memory
  • 🎓 Students learning about AI systems

If you're building anything with AI that needs to remember stuff (spoiler: that's everything), you're in the right place.

What You'll Learn

The curriculum, but make it fun 📖

This documentation will teach you:

  1. Core Concepts - Workspaces, Bots, Memories (oh my!)
  2. Quick Start - Get running in 5 minutes
  3. SDK Deep Dives - Python, JavaScript, you name it
  4. Best Practices - How to memory like a pro
  5. Advanced Patterns - For when you're feeling fancy

Prerequisites

The boring but necessary part 📋

What You Need:

  • ✅ An API token from memof.ai
  • ✅ Basic programming knowledge (any language)
  • ✅ Coffee (optional but recommended ☕)

What You Don't Need:

  • ❌ A PhD in computer science
  • ❌ Years of AI experience
  • ❌ Patience (we keep it quick!)

Your First Win

Let's get you a quick victory. Here's memory in 10 lines:

from memofai import create_moa_client

client = create_moa_client(api_token='moa_your_token')
workspace = client.workspaces.create({'name': 'My First App'})
bot = client.bots.create({'name': 'Helper', 'workspace_id': workspace['id']})
client.memories.create({'bot_id': bot['id'], 'content': 'User loves quick wins'})
results = client.memories.search({'bot_id': bot['id'], 'query': 'user preferences'})

print(f"Found: {results[0]['content']}")
# Output: Found: User loves quick wins

Done. You just gave an AI permanent memory. Feel powerful yet?

What's Next?

Choose your own adventure 🗺️

Questions?

We love questions! 💬

The Promise

Our commitment to you 🤝

  1. Simple: If it feels complicated, we failed
  2. Fast: Your time is valuable
  3. Reliable: Your data matters
  4. Fun: Docs don't have to be boring

Welcome aboard! Let's build something that actually remembers things. 🎉

Ready to get started? → Core Concepts | Quick Start

ON THIS PAGE