🧠 AI Memory Backend

A minimal, production-style backend for building AI systems with persistent memory.

This service lets you store, search, and recall user-specific memories — and generate structured context for LLMs.

---

🚀 Features

• Save structured user memories
• Search memories with relevance scoring
• Recall memories by topic
• Automatic topic inference
• Context generation for AI conversations
• Lightweight FastAPI + SQLite setup

---

📦 API Overview

• POST /memory/save — store a memory
• POST /memory/search — search memories by query
• POST /memory/recall — recall top memories
• POST /chat/context — build memory-aware context for AI
• GET /memory/{memory_id} — fetch a memory
• GET /health — health check

---

⚡ Quick Start

git clone <your-repo>
cd ai-memory-backend

python3 -m venv venv
source venv/bin/activate

pip install -r requirements.txt
uvicorn server:app --reload

Open: http://127.0.0.1:8000/docs

---

🧠 Example

Save a memory:

{
  "user_id": "user_123",
  "content": "My favorite food is sushi",
  "importance": 4
}

Search memories:

{
  "user_id": "user_123",
  "query": "what do I like to eat?"
}

---

🎯 Use Cases

• AI assistants with long-term memory
• Personal AI companions
• Memory-augmented chatbots
• RAG-style systems without heavy infrastructure
• Prototyping AI memory systems quickly

---

🧩 Notes

• Uses SQLite by default (easy local setup)
• Can be swapped for Postgres / vector DB later
• Designed to be simple, hackable, and extensible

---

⭐️

If this helped you, consider starring the repo!