TeleScribe | Devpost

poster

Inspiration

Many mobile company customers encounter errors and problems on their smartphones on a daily basis. Navigating the error, such as locating the error code or name can be a hassle, not to mention the millions of google results with different trouble shooting solution. We decided to fix this issue by having the user simply be able to take a screenshot of the error pop-up and describe it in a sentence or two, and our app will be able to give step by step troubleshooting directions to fix the problem.

What it does

A user who encounters a phone error anywhere from a simple text messages not forwarding correctly to a complex APN setup issue, the user can simply take a screenshot of the error/issue and describe it on the app. The AI takes that image and identifies important error codes and technical terms, by training this model it will become more accurate. The text is sent to our server handler that forwards the message to azure cognitive text API to get the sentiment of the user's description while highlighting the main phrases in the sentence. Once this information is available it is compared with stored troubleshooting procedures from T-mobile's troubleshooting website. The procedure is then forwarded back to the user on the app. The application is able to take any language as Microsoft's translator was used to determine the language and automatically translate it to English.

How we built it

Hard work and perseverance! Microsoft Azure, for all text analysis and translation. Android developer to make the application. T-mobile store locator to find the nearest stores to the customer. Node.js to make everything communicate.