Skip to main content

Welcome to JARVIS

JARVIS is a real-time person intelligence platform that combines facial recognition, autonomous web scraping, and AI-powered research to deliver comprehensive dossiers on individuals in seconds. Built for the Browser Use + YC Web Agents hackathon, JARVIS uses Meta Ray-Ban smart glasses to capture photos, identify individuals, and orchestrate a swarm of browser agents to gather intelligence across the web.

Key Features

Real-Time Identification

Facial recognition via Meta Ray-Ban smart glasses with PimEyes integration and AI vision

Agent Swarm Research

Multi-agent OSINT orchestration scraping LinkedIn, Twitter, Instagram, and Google

Live Streaming UI

COD-style corkboard interface with animated dossier cards and real-time updates

AI Synthesis

LLM-powered intelligence aggregation and report generation

How It Works

1

Capture

Use Meta Ray-Ban smart glasses or phone camera to capture a photo of a person
2

Identify

Facial detection and reverse image search via PimEyes identifies the person
3

Research

Browser Use agent swarm autonomously scrapes LinkedIn, Twitter, Instagram, and Google
4

Synthesize

AI aggregates all gathered intelligence into a comprehensive dossier
5

Stream

Results stream in real-time to the corkboard UI as data arrives

Use Cases

JARVIS is designed for:
  • Founders at networking events identifying potential investors, partners, or hires
  • Sales professionals at conferences qualifying leads in real-time
  • Security teams performing rapid OSINT research
  • Anyone wanting ambient intelligence about people they encounter
JARVIS was built for the Browser Use + YC Web Agents hackathon as a demonstration of multi-agent research orchestration. It showcases the power of autonomous browser agents for real-time intelligence gathering.

Get Started

Quickstart

Get JARVIS running in under 15 minutes

Installation

Complete installation and setup guide

Architecture

Understand JARVIS’s system design

API Reference

Explore the API endpoints

Tech Stack

JARVIS is built with:
  • Backend: Python FastAPI with Browser Use SDK for agent orchestration
  • Frontend: Next.js with Framer Motion for animated UI
  • Database: Convex (real-time) + MongoDB (persistent storage)
  • AI/ML: GPT-4o (vision), Gemini (synthesis), MediaPipe (face detection)
  • Research APIs: PimEyes (facial recognition), Exa (structured research)
  • Observability: Laminar tracing for accuracy verification
JARVIS is a hackathon demonstration project. It performs automated OSINT research and should be used responsibly and in compliance with applicable laws and terms of service.

Build docs developers (and LLMs) love