MemoriaAI - Private Assistant with Memory & Tools

Inspiration

Modern AI models are powerful, but they reset after every conversation and lack continuous user context. We created MemoriAI to bridge this gap by building a private, context-aware AI assistant equipped with persistent memory and reusable tool execution.

What it does

MemoriAI serves as a long-term personal co-pilot:

• Persistent Context & Memory: Remembers core user preferences, active projects, and past notes across independent sessions using a vector storage layer.
• Dynamic Tool Execution: Leverages custom API tools for live web queries, structured document summaries, and code analysis.
• Privacy Controls: Gives users granular control over saved memory vectors, allowing them to inspect or clear stored context at any time.

How we built it

• Backend Infrastructure: Powered by Nebius AI Cloud / Token Factory running NVIDIA open-source models (Nemotron-4 340B) for high-speed LLM inference.
• Testing & Reliability: Built with pytest and automated using GitHub Actions CI/CD to ensure continuous integration stability.

Built With

• chromadb
• css
• html
• javascript
• langchain
• llms
• nebius-ai-cloud
• nvidia
• python
• web-apis