Show HN: I Developed AI Memory Booster: Self-Hosted AI with Long Term Memory

3 points

a year ago

I recently developed and open-sourced a project called AI Memory Booster, which combines Ollama (for running local LLaMA models) with ChromaDB to give AI systems persistent memory across sessions.

The project is fully self-hosted and privacy-first — everything runs locally via a Node.js API, a simple React UI, and Docker support for easy deployment.

Key Features: 1. Ollama-powered inference (LLaMA 3.2 and other models). 2. Persistent memory via ChromaDB (store and recall data across sessions). 3. Works on CPU or GPU, tested on local laptops and free-tier cloud VMs. 4. API-first approach with /learn and /recall endpoints. 5. Ready-to-use React web interface + install.sh script for fast setup.

Use Cases: 1. Build a local AI chatbot with memory. 2. Power a self-hosted assistant that remembers conversations or tasks. 3. Add a memory layer to Ollama agents or automation workflows. 4. Integrate into existing Node.js applications.

The source code is now available on Github: https://github.com/aotol/ai-memory-booster

I’d love feedback from the community — especially ideas on improving long-term memory handling or other integrations you’d find useful!