ProtoBuddy

Inspiration we're creating our own electrical circuits, and we kept finding the same error every single time we built one. We didn't understand which two components were compatible, and to find that out, we used to read multiple data sheets. It would take us a long time to find out which components were compatible and how to make a circuit.

What it does

ProtoBuddy is an intelligent AI agent that eliminates hardware component selection guesswork by providing accurate, datasheet-verified compatibility recommendations. Unlike existing LLMs that hallucinate on technical specs, ProtoBuddy actively scrapes datasheets, builds a structured knowledge base with RAG (Retrieval-Augmented Generation), and provides grounded answers with exact specifications and wiring diagrams.

How we built it

How we built it We built ProtoBuddy using three integrated tools: Apify for dynamic datasheet scraping and structured data extraction, Redis for our RAG-powered knowledge base that stores component specifications with confidence levels, and Claude API for intelligent compatibility reasoning and algorithmic checks. The system parses PDFs into structured ontologies, runs comprehensive compatibility algorithms for voltage, current, protocol, and pin requirements, then generates actionable solutions with exact wiring diagrams and component values.

Challenges we ran into The main challenge was achieving zero hallucinations while maintaining fast response times, requiring us to build sophisticated datasheet parsing with confidence scoring and safety-first uncertainty flagging. We also struggled with inconsistent PDF datasheet formats across manufacturers, necessitating robust extraction algorithms that could handle varied layouts while maintaining accuracy.

Accomplishments that we're proud of We successfully created an AI system that achieves zero specification hallucinations by tracing every answer back to actual datasheets, while detecting 95% of compatibility issues before prototyping. Our system provides actionable solutions like exact resistor values for voltage dividers and specific I2C multiplexer recommendations, going beyond simple compatibility checking to generate complete wiring diagrams and power budget calculations.

What we learned We learned that domain-specialized AI with curated knowledge bases significantly outperforms generic models for technical applications, and that confidence scoring is essential for safety-critical engineering decisions. The importance of structured data extraction from unstructured PDFs became clear, as did the value of progressive knowledge base building where each query improves the system's accuracy.

What's next for ProtoBuddy Next, we'll expand Phase 2 features including progressive knowledge base building with cross-referenced specifications, intelligent caching strategies, and environmental compatibility checks for temperature and humidity ranges. We plan to add BOM generation with purchase links, Arduino/MicroPython library verification, and support for PCB design integration to serve our target market of 50 million engineers worldwide.

Built With

• anthropic
• apify
• lovable