Nabu- Always connected

Nabu is an AI companion that "sees" through Gemini and "speaks" with empathetic, cloned voices via ElevenLabs. We turn cold technology into a familiar presence to end elderly isolation.

Comment

Inspiration

Nabu was born to make technology accesible for everyone, creating an AI companion that feels human .

What it does

A multimodal AI assistant for technical support featuring voice interaction (ElevenLabs), real-time visual aid, and security analysis to detect fraud.

How we built it

Nabu was developed using a modular architecture that synchronizes real-time data processing with a high-end user interface, centering its core on a Node.js server deployed on a Digital Ocean Droplet to ensure low latency and multi-platform stability. The system orchestrates several key APIs, including a Generative AI engine for natural language processing and fraud detection, alongside the ElevenLabs API to provide empathetic, human-like voice synthesis. Additionally, we leveraged the browser's native MediaDevices API to allow the AI to "observe" the user's screen and provide precise visual assistance. This entire ecosystem is delivered through a professional frontend built with HTML5, JavaScript, and CSS3.

Challenges we ran into

Implementing a robust authentication system that remains consistent across different platforms while deployed on Digital Ocean was a major hurdle.

Accomplishments that we're proud of

Building a high-end, native-feeling interface from scratch in hours and ensuring a seamless, low-latency bridge between the frontend and the AI backend.

What we learned

The power of "look and feel": how professional CSS and proper deployment architecture can instantly build user trust. We also sharpened our real-time debugging and rapid UI iteration skills.

What's next for Nabu

Implementing real-time screen analysis for step-by-step guidance and adding emotional intelligence to the avatar for a more empathetic user experience.

Share this project:

Updates