Fintech
Travel Tech
Electric Vehicle
Health Tech
Edtech
IT
Logistics
Retail
Ecommerce
Startup Ecosystem
Enterprise Tech
Clean Tech
Consumer Internet
Agritech
TBO Tek has automated 70% of special travel requests using AI voice bots, significantly reducing response time, TBO Tek's customer experience VP Garima Pant said at Inc42’s 'The GenAI Summit'
Cars24's VP of engineering Deepak Gupta recommended building in-house AI solutions for complex use cases with rich context, while simpler problems can be effectively addressed with outsourced NLP-based solutions
Around 60% of clients have moved to GenAI with consistent accuracy rates of around 89%, with companies focusing more on providing better context to foundation models than fine-tuning, said CoRover.ai's Ankush Sabbharwal
Travel tech company TBO Tek has been leveraging GenAI for automation and about 70% of special requests from travel agents are now being handled by AI voice bots, the company’s customer experience VP Garima Pant said.
“We are able to cater to almost 70% of special requests globally through voice bots,” Pant said during a panel discussion at Inc42’s ‘The GenAI Summit’ today.
She added that AI has significantly reduced response time for special requests from travel agents. Earlier, such requests required a person to make amendments to the system but are now completely handled using GenAI. The range and complexity of the requests that the system can handle has multiplied.
OfBusiness cofounder and CBO Nitin Jain, CoRover.ai founder and CEO Ankush Sabharwal, Swiggy’s VP of data science and analytics Goda Ramkumar, and Cars24’s VP of engineering Deepak Gupta were among the other members of the panel, who held discussions on the topic, ‘Riding The Conversational AI Wave: What’s Next?’.
Should Conversational AI Be Built In-House Or Outsourced?
Giving her views on the topic of building or outsourcing AI solutions, Gupta suggested that simpler use cases can be outsourced. “If the problem statement is not really very complex, NLP (natural language processing)-based things can work it out,” he said.
For complex implementations, he recommended developing solutions in-house. “If you are really very context rich and have very complex use cases of intent and entities, then (considering) the way you know your business… it makes (sense to) build your own solution,” he said.
Gupta revealed that when Cars24 made LLM-based chatbots one-and-a-half year ago, the bots achieved only 50% accuracy with NLP-based solutions due to language variations in customer queries.
Sharing adoption numbers, Sabharwal, whose company builds conversational AI solutions, said, “60% of our clients have moved to GenAI and 40% are still with classic NLP intent classification and entity extraction.”
Despite the shift to GenAI, Sabharwal noted that accuracy rates have remained consistent. “The accuracy did not change from earlier classic NLP to GenAI. It’s been consistent… 89% average accuracy,” he added.
Meanwhile, Swiggy’s Goda highlighted the need for voice interfaces, especially for delivery partners. “For them, it should be much more voice based because you are on the road, you are driving,” she said.
The panel discussion also revealed that companies are moving away from fine-tuning models and are instead focussing on providing better context to foundational models for more effective AI implementation.
With the adoption of GenAI on the rise across industries, India’s GenAI market is expected to become an over $17 Bn opportunity by 2030.
