Inspiration
Fixen was inspired by the constant frustrations homeowners and renters face when something breaks, leaks, or stops working — and they don’t know where to start. Many people resort to costly repair visits or unverified DIY advice online. We thought, what if anyone could access a trusted, conversational repair assistant powered by AI?
What it does
- Diagnostic Guidance: Users describe their issue in plain language through an intuitive chat interface.
- Severity Levels: Issues are categorized by severity (e.g., minor wall cracks vs. major structural damage).
- Interactive Q&A: The assistant asks follow-up questions to narrow down possible causes.
- Actionable Recommendations: Provides step-by-step troubleshooting tips or suggests professional help when necessary.
- History & Logs: Tracks past diagnostics so users can revisit previous issues.
How we built it
- Frontend:
- Scaffolded quickly with Bolt.new
- Designed a clean, friendly UI in Figma
- Scaffolded quickly with Bolt.new
- Backend:
- Used Supabase to store user logs and diagnostic histories
- Used Supabase to store user logs and diagnostic histories
- AI Assistant:
- Powered by ChatGPT for natural-language understanding
- Supplemented with Claude- and Perplexity-inspired logic for real-time validation
- Powered by ChatGPT for natural-language understanding
- Conversational Flows:
- Mapped out diagnostic trees for common household categories
- Conducted extensive user testing to refine question sequences
- Mapped out diagnostic trees for common household categories
Challenges we ran into
- Language Safety: Fine-tuning the assistant’s wording to avoid incorrect or potentially dangerous instructions.
- Context Memory: Integrating session-to-session memory without compromising user privacy.
- UX Balance: Designing an interface that feels helpful and supportive, yet not overwhelming or too technical.
Accomplishments that we’re proud of
- Created a fully functional MVP with dynamic diagnostics
- Designed and tested unique flows for wall repair scans
- Established a clean brand identity and user interface tailored to our target audience
What we learned
- How to balance UX and technical feasibility in AI-first applications
- How to validate AI responses with fallback systems like Perplexity for fact-checking
- How to manage cross-functional team roles in a fast-paced project (design, development, testing)
What’s next for Fixen
- Add voice assistant functionality using Whisper or another voice-to-text model
- Expand diagnostics into HVAC, roofing, and car repairs
- Partner with local professionals for monetized referrals
- Release a beta to real users for feedback and iteration
Built With
- bolt.new
- chatgpt
- figma
- netlify
- openai
- perplexity
- react
- supabase




Log in or sign up for Devpost to join the conversation.