HVAC AI Assistant

HVAC companies lose thousands of dollars every week in missed after hours calls and leads. With 105,000+ contractors in the US and only 15% having any form of digital customer service, the gap is massive. Existing solutions like Service Titan and Intercom are either too expensive or too generic for the trade. HVAC AI Assistant was built to solve this problem a domain-specific AI chatbot that knows HVAC inside and out, handles customer questions 24/7, and captures leads automatically so no opportunity is ever missed. IBM Bob was the core development partner throughout the entire build. Bob read the complete repository, explained the architecture, wrote the HVAC system prompt with South Florida specific knowledge including humidity, salt air, and hurricane preparedness, wired in prompt caching for cost efficiency, added lead capture logic, reviewed the codebase for production issues, and taught the underlying concepts at every step. All 14 Bob prompt sessions are fully documented in the bob_sessions folder of the GitHub repository with screenshots, markdown exports, and video recordings showing Bob in action as a true development partner. The technical stack includes Python Flask backend, vanilla JS frontend with real time SSE streaming, Claude Sonnet with prompt caching reducing API costs by up to 90%, and a fictional HVAC FAQ dataset integrated as the knowledge base. The application is deployed live on Railway and ready for judges to interact with. The business model is SaaS at $99-$299 per month per company 100 clients in South Florida alone represents $239K ARR. Phase 2 adds SMS and WhatsApp. Phase 3 brings AI to technicians in the field. This is how IBM Bob turns an idea into impact faster.