Assisto Technologies works with IBM to bring multilingual voice bots for enhanced customer experience

By Yukti Panjabi and Sakshi Jain

In today’s digital environment, customer conversations often shift fluidly between English and regional languages – a practice known as “code-mixing.” For organisations operating in linguistically diverse regions, being able to understand and respond to such mixed-language conversations has become increasingly important for delivering effective support and building long-term relationships.

Noida-based Assisto Technologies, founded in 2021, identified this challenge early and has been working to address it through artificial intelligence-powered voice technologies.

Developing voice AI for real-world communication

Assisto specialises in Natural Language Understanding (NLU) and Natural Language Generation (NLG), helping enterprises implement conversational AI solutions that enable real-time transcription, speech-speech translations, sentiment analysis, and contextual responses. Its core offering, iAssist, is designed to support multilingual and emotion-aware voice interactions.

To enhance the platform’s capabilities, Assisto teamed up with IBM to integrate watsonx.ai and Granite large language models (LLMs), leveraging these tools to handle complex, code-mixed language inputs such as Hinglish.

“We worked with IBM to combine our domain expertise in conversational AI with their foundation models and analytics technologies,” said Sakshi Jain, Sales Director, Assisto Technologies Pvt. Ltd. “The result is a more human-like voice bot experience that can understand and respond across multiple languages.”

The solution uses Assisto’s Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) systems to transcribe multilingual speech accurately. IBM’s watsonx.ai enables analysis of tone, sentiment, and emotion in real time, while the Granite LLMs, fine-tuned by Assisto, improve response accuracy in code-mixed conversations.

A reporting layer built with IBM Cognos brings these insights together in a single dashboard, helping support teams access call summaries, sentiment trends, recurring issues, and performance metrics more easily.

Real-World Outcomes for Businesses and Measurable Benefits

– 60-70% reduction in time spent on manual transcription

– Up to 50% savings in operational costs via automation

– 20-30% improvement in customer satisfaction scores

– 15-20% increase in customer retention and engagement

By making real-time insights more accessible, the platform supports quicker decision-making and more personalised customer service.

Enabling scalable and inclusive engagement

“The combination of IBM watsonx and Granite LLMs with Assisto’s voice AI platform has enabled a practical approach to solving the code-mixed language challenge,” said Yukti Punjabi, Director, Ecosystem, IBM India & South Asia. “This collaboration helps organisations deliver more inclusive, emotionally aware, and efficient customer interactions.”

The solution is now available across markets in the US, Europe, Asia-Pacific, and other regions and is being deployed in sectors such as banking, telecom, insurance, and retail.

Looking ahead, Assisto plans to expand support for more regional languages, enhance emotional context recognition, and continue developing tools that help businesses understand their customers better across voice-based touchpoints.