About Trace

Inspiration

Appliance-service requests are often fragmented. Customers repeatedly explain the same issue, while technicians arrive without enough information. We built Trace, an AI-powered service agent that manages the journey from customer report to resolution.

What It Does

The customer enters the product serial number, and the agent asks relevant questions based on the appliance. It collects symptoms and evidence, creates a Freshdesk ticket, assigns the relevant technician, and notifies the customer through WhatsApp.

The technician receives the complete context and a guided inspection plan. As findings are reported, the agent adapts the remaining tasks. After technician confirmation, it records the result and closes the Freshdesk ticket.

How We Built It

We used React, Vite, Node.js, Express, OpenAI, Freshdesk, Twilio WhatsApp, and Vercel. OpenAI powers the adaptive conversation and voice prompts, while Freshdesk and Twilio enable real business actions.

Challenges and Learning

Our main challenge was making the experience genuinely agentic instead of creating another chatbot. We focused on contextual questioning, tool-driven actions, adaptive technician guidance, and human confirmation. We learned that effective agents must not only respond. They must reason, act, adapt, and complete workflows.

Built With

• express.js
• freshdesk
• node.js
• openai
• react
• twilio
• vite