The latest articles on DEV Community by Akshit Mishra (@akshit_mishra_30f3e40aa16). https://dev.to/akshit_mishra_30f3e40aa16 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4226075%2F87f3f33a-d9f8-4cbc-8097-6016130ee338.png https://dev.to/akshit_mishra_30f3e40aa16 en Akshit Mishra Wed, 04 Nov 2026 11:30:12 +0000 https://dev.to/akshit_mishra_30f3e40aa16/building-pratinidhi-a-multilingual-assistant-for-navigating-native-language-speakers-through-4875 https://dev.to/akshit_mishra_30f3e40aa16/building-pratinidhi-a-multilingual-assistant-for-navigating-native-language-speakers-through-4875 <p>How can someone explain a customer-care problem when the helpline doesn’t offer their language?</p> <p>We explored this through Pratinidhi, our prototype for the Great Indian AI Internship Challenge.</p> <p>The demonstration follows this journey:</p> <ol> <li>A customer explains her concern in Marathi.</li> <li>Pratinidhi identifies missing details and asks for them by voice.</li> <li>She gives permission before the case is shared.</li> <li>The representative handles English support while she follows the conversation in Marathi.</li> <li>When additional information is needed, it asks her privately, confirms the answer and continues. We used Gnani Prisma for speech recognition and Timbre for speech synthesis.</li> </ol> <p>The current demonstration combines recorded, genuine speech API outputs with a simulated support call. It explores the interaction design before live telephony integration.<br> <a href="https://lnkd.in/p/gh7hAUbD" rel="noopener noreferrer"></a></p> <p>Which failure mode would you prioritise before taking this into real calls?</p> a11y ai nlp