MY-CALL-E-PROJECT

I built MY-CALL-E-PROJECT to make finding repair help less time-consuming. When someone needs a plumber or another service provider, they may have to make several calls and repeat the same questions about availability, charges, and arrival times. I wanted to explore how an AI calling agent could help with this task while keeping the user in control. The application collects the service request, location, preferred time, budget, and provider’s phone number. It prepares a call plan and asks the user to approve it before making an outbound call through CALL-E. The agent asks for service information without making a binding booking or agreeing to charges. Afterwards, the application displays the call status, summary, transcript, and structured results. I built MY-CALL-E-PROJECT using Node.js, JavaScript, HTML, CSS, the CALL-E SDK, and SQLite. The backend handles call requests and checks their progress, while SQLite stores call records and workflow information. The CALL-E API key remains on the server. I also added confirmation controls, request identifiers to help prevent duplicate calls, and audit logging. The main challenge was understanding why my initial calls failed. The application initially displayed a general error, so I added diagnostic logging to reveal the underlying response. This showed that the calling setup did not support the Pakistan-and-English combination I was testing. The organizers then provided an official US testing hotline, which allowed me to continue without requiring a personal number in a supported country. My proudest moment was completing a real call to that hotline and listening to its recording on the CALL-E website. The application also returned the conversation transcript, summary, and structured results. Since the hotline could not provide actual repair appointments or prices, the results correctly marked those details as unknown. This demonstrated the calling workflow and its handling of missing information, although testing with real service providers remains a future step. Through this project, I learned how to connect a web application to a voice-agent service, track calls over time, and turn conversations into useful structured information. I also learned that a completed call does not always mean the user’s task has been completed; clear reporting of uncertainty is essential. Next, I plan to simplify the activity display, make supported regions and languages easier to select, and improve the presentation of call outcomes. I would also like to support comparisons across multiple approved providers and test the application with real repair businesses. My goal is to make MY-CALL-E-PROJECT a practical assistant that helps people gather information and decide whom to contact next.