One of the leading conversational AI companies, Jio Haptik, is using Microsoft Azure Cognitive Services to enhance existing Hindi conversational AI models’ accuracy.
This popular AI translation model that enables end-to-end conversations in Hindi, English, and Hinglish is integral to allowing users to interact with the Intelligent Virtual Assistants at Jio Mobility.
The one-of-its-kind model has helped the Jio Mobility team provide support across multiple categories by enhancing local language queries by 2.5 times and reducing human interventions by 80%.
Read More: US Officials Order Nvidia To Stop Exporting Computing Chips For AI Work To China
With Jio Mobility, Haptik provides chatbots across three platforms: WhatsApp Business account, MyJio App, and Jio website. This solution has assisted Jio Mobility in engaging in 2M+ conversations from customers in Hindi and Hinglish with an 80% decrease in human interventions and a 2.5 times increase in localized queries.
Microsoft, in collaboration with Jio Haptik, facilitates vernacular support for users by leveraging transfer learning from multilingual and monolingual data to create this unique language model. Their unified efforts have helped Jio Mobility’s customer care feature a one-of-a-kind conversational service in Hindi and Hinglish.
Basic translation models have existed in the market for a long time. Still, they are not equipped to handle grammatical errors, and unstructured sentences are essential for conversational agents to perform well. However, this one-stop solution by Azure Cognitive Services and Jio Haptik manages Hindi in Roman script (Hinglish) and Devanagari, allowing users to switch between languages while functioning at high accuracy for short sentences and domain-specific conversations.