Fixit: AI-Powered Real-Time Appliance Troubleshooting

Inspiration

Long customer service wait times and complex troubleshooting guides make appliance issues frustrating. We wanted to create an AI-powered assistant that provides instant solutions, eliminating the need to contact customer support or sift through lengthy manuals.

What it does

Fixit is a real-time appliance troubleshooting agent that:

• Accepts image, text, audio, or video as input
  
• Identifies the appliance type, brand, model, and other key details
  
• Finds official troubleshooting guides online
  
• Extracts relevant information using AgentQL
  
• Synthesizes and delivers clear, AI-powered solutions
  

How we built it

We leveraged Agno’s multi-agent architecture and Weave from Weights & Biases for efficient task execution. Our backend is built with:

• Python & Flask for handling requests
  
• AgentQL for web scraping
  
• LlamaIndex for information synthesis
  

The frontend uses:

• Next.js for an interactive UI
  
• Stytch for authentication
  

Challenges we ran into

• Processing multi-modal inputs efficiently
  
• Seamless integration between agents and external APIs
  

Accomplishments that we're proud of

• Successfully built a fully automated troubleshooting system
  
• Optimized AI agent workflows using Weave
  
• Integrated multi-modal input processing for user flexibility
  

What we learned

• Best practices for multi-agent coordination
  
• Effective ways to integrate real-time web scraping
  
• Optimizing AI models for knowledge extraction
  

What's next for Fixit

• Voice-over assistance with video troubleshooting
  

🚀 Fixit is just getting started!

Built With

• agentql
• agno
• flask
• llamaindex
• next.js
• openai
• python
• stytch
• weave