Skip to content
Finalist

DrystAI

Never forget a face or conversation again.

Code Video Cerebral Valley

Core trick

Matches webcam frames against a stored face database via a vision model, then overlays contact notes on a live feed.

Tags

visionvoice

Description

Never forget a face or conversation again. DrystAI ("Dryst" means vision in Sanskrit), turns your smart glasses into a real-world memory. It recognizes faces, pulls up past conversations, and shows you names, companies, and context right when you need it. No awkward pauses. No “sorry, what was your name again.” It’s LinkedIn, your notes, and your brain working together in real life. Built for people who actually want to remember.
RadonUmar/drystai★ 10 forksTypeScriptCreated Oct 19, 2025 · last push Oct 19, 2025

README excerpt

# DrystAI

> Never forget a face or conversation again.

![DrystAI Demo](public/demo.png)

*Live camera feed with AI-powered facial recognition, real-time transcription, and AR-style overlays*

## Overview

DrystAI is an AI-powered networking assistant that provides real-time facial recognition, conversation tracking, and intelligent summarization. The system captures webcam screenshots, analyzes faces using vision AI, transcribes conversations via speech recognition, and retrieves contextual information about people you've met before.

## Architecture

### Core Technologies

**Frontend**
- Next.js 15 (App Router)
- TypeScript
- Tailwind CSS (Glass morphism UI)
- WebRTC for camera access
- Web Speech API for real-time transcription

**Backend**
- Next.js API Routes (serverless functions)
- MongoDB Atlas (vector search enabled)
- Google Gemini AI (vision, embeddings, NLP)
- Fetch.ai uAgents (web search integration)

**AI/ML Pipeline**
- Gemini 2.5 Flash for multimodal AI tasks
- text-embedding-004 for vector embeddings (768 dimensions)
- MongoDB Atlas Vector Search for similarity matching

## How It Works

### 1. Face Recognition System

**Input**: Webcam screenshot (JPEG image)

**Process**:
1. Image sent to Gemini Vision API (`gemini-2.5-flash`)
2. AI generates detailed 10-point facial description (structure, features, expressions, distinguishing marks)
3. Description converted to 768-dimensional vector using `text-embedding-004`
4. Vector stored in MongoDB with person metad

Team