This article was created for the purposes of entering the Gemini Live Agent Challenge hackathon. #GeminiLiveAgentChallenge
$44 Billion in Electronic Waste. Zero AI Tools to Fix It.
Every year, 44 billion dollars worth of electronics gets thrown away — not because it's broken beyond repair, but because nobody knows what's wrong with it.
A technician stares at a PCB. Is that capacitor bulging? Are those solder joints cold? Which IC is this? Where's the datasheet?
The tools haven't changed in decades: a multimeter, a magnifying glass, and Google. Meanwhile, AI can drive cars and generate movies — but it can't look at a circuit board and tell you what's wrong?
Until now.
Meet CircuitSense
CircuitSense is an AI lab partner that sees your hardware through your camera and talks to you about it in real-time.
No typing. No uploading photos. No waiting.
Point. Ask. Fix.
🔍 SCAN — Identifies every component, checks physical condition, detects protocols
🔧 DIAGNOSE — Hunts for faults: burnt traces, cold solder, bent pins, swollen caps
🔌 WIRE — Shows you exactly how to connect components with ASCII wiring diagrams
📚 LEARN — Explains every component in beginner-friendly language
Why This Matters Now
The maker movement has exploded. There are 50 million Arduino boards in the wild. ESP32 sells millions per month. But the learning curve is brutal — especially for students and hobbyists who don't have a mentor sitting next to them.
CircuitSense IS that mentor.
It sees your ESP32 and knows the pinout. It spots the bent header pin you missed. It tells you GPIO21 is SDA and GPIO22 is SCL before you even ask. It warns you when you're about to connect 5V to a 3.3V input.
This isn't a chatbot that answers electronics questions. It's an AI that looks at YOUR specific hardware and gives YOU specific answers.
The Secret: Two Brains Are Better Than One
Most AI assistants use one model for everything. CircuitSense uses two — simultaneously.
Brain 1 talks to you. Real-time voice conversation powered by Gemini Live API. It hears your questions, sees your camera, and responds naturally.
Brain 2 analyzes silently. Every 15 seconds, it scans your camera feed and produces a structured component inventory with health assessments, protocol detection, datasheet links, and project suggestions. This populates a rich sidebar that updates continuously.
The result: you never stare at a blank screen. Even while Brain 1 is thinking about your complex wiring question, Brain 2 has already identified 15 components and flagged a suspect capacitor.
What CircuitSense Sees
In testing, CircuitSense successfully identified and assessed:
- Development boards — ESP32, Arduino, STM32, NodeMCU
- Sensors — MPU6050 IMU, HC-SR04 ultrasonic, LD2410C radar
- Complex PCBs — set-top boxes, control boards, custom designs
- Individual faults — bent pins, oxidized contacts, ribbon cable damage
- Protocols — I2C, SPI, UART, USB, HDMI, Ethernet, Bluetooth
It generates real-time health scores, wiring diagrams, and even suggests what projects you could build with the components it sees.
The Modes That Make It Useful
CircuitSense isn't one tool — it's four:
SCAN mode is your magnifying glass on steroids. Point at any board, get a full component inventory in seconds.
DIAGNOSE mode is your quality inspector. It specifically looks for physical damage, corrosion, cold solder joints, and manufacturing defects.
WIRE mode is your wiring assistant. Ask "how do I connect this ESP32 to this MPU6050?" and get an ASCII diagram you can follow with one hand while holding a soldering iron with the other.
LEARN mode is your patient tutor. It explains what every component does, how they work together, and what each pin is for — in language a beginner can understand.
Who Needs This
- Electronics students learning to identify components and debug circuits
- Hobbyists and makers building projects with Arduino, ESP32, Raspberry Pi
- Repair technicians diagnosing faults in consumer electronics
- QA engineers performing visual inspection of PCB assemblies
- Anyone who's ever stared at a circuit board and thought "what IS that?"
Built With
- Gemini 2.5 Flash Native Audio (Live API) for real-time voice
- Gemini 2.5 Flash for structured vision analysis
- Google Cloud Run for hosting
- Python, FastAPI, WebSocket architecture
Try It Right Now
Live Demo: CircuitSense on Cloud Run
Source Code: github.com/kryptonus/CircuitSense
Grab any electronics — an Arduino, a phone charger PCB, even a TV remote board — point your camera at it, and watch CircuitSense break it down in real-time.
The age of AI-powered electronics repair isn't coming. It's here.
Built for the Gemini Live Agent Challenge. #GeminiLiveAgentChallenge
Top comments (0)