Startups chase finserv companies seeking GenAI customer support use cases

Time of India

11-07-2025

Academy
Empower your mind, elevate your skills
ETtech
The highly regulated financial services industry is finding application of generative artificial intelligence models in customer-facing operations, from onboarding to servicing and basic query resolutions, as adoption of the technology goes up and its cost comes down.Large banks spend heavily on running contact centres or backend support teams with human agents to service their clients. Now with agentic AI, a large part of the basic queries can be solved by software, say industry insiders.'Our models can reduce the cost for financial services by 40-50%, manage a lot of routine things,' said Gnani AI chief executive Ganesh Gopalan. 'If our client is serving 50,000 customers today, tomorrow they can scale it by five times, instead of hiring more agents to do this work.'Bengaluru-based Gnani AI, backed by InfoEdge Ventures, builds conversational AI tools for finserv companies.Startups like Gnani.ai, Arya.ai, Yellow.ai and Hyperface are building AI use cases targeted at different service categories. While some models have already gone live, a few are under deployment.'Currently in India the cost of running a GenAI model is at par with contact centre employees. But in the future, AI models can be scaled up exponentially which is not the case of call centres, and that is the big benefit that companies are looking for,' said Ramanthan RV, chief executive at Hyperface, which powers credit card stacks for banks.The model Hyperface has built is already deployed at a fintech startup, assisting customers who have downloaded their application but haven't begun using the app yet. It is also deploying another model with a bank in the credit card issuance business.Voice bots and chatbots have been around for many years already, but with GenAI capabilities these bots are becoming smarter, more advanced and capable of speaking in a proper human voice.A top executive at a fintech startup explained that most of these first-generation bots were trained through machine learning models which have now gone obsolete. Currently banks are taking enterprise solutions from the likes of OpenAI and Microsoft and training them with specific capabilities. The bots can then not only answer basic queries but also solve problems on the fly.'Financial services is a major sector for us; the use cases are majorly in card issue support, loan processing, customer acquisition and information regarding claims,' said Rashid Khan, cofounder of Yellow AI, which is backed by the likes of Lightspeed and Westbridge Capital. 'This AI model focuses on accuracy, latency and making the bot sound more human,' he said.Even large financial firms are taking note of the power that these new technologies can bring to their business. Bajaj Finance , one of the largest non-banking lenders in the country, acquired a 12% stake in Protectt.ai, a mobile security firm. In April 2024, listed banking technology company Aurionpro acquired Arya.ai to push AI models to its banking clients.India currently has around 44 GenAI companies in speech and voice recognition, including Sarvam, Uniphore and Augnito, and 25 of them are funded, according to a Tracxn report published in June. While startups are enthusiastic, the government is encouraging and venture funds have taken note of the opportunity as well, integration with legacy banking software systems and stringent regulations make the deployments tricky.'Regulatory compliance remains the primary challenge. Every solution must navigate strict BFSI data security requirements, with companies needing ISO 27001 and GDPR compliance. The regulatory framework for AI in collections is still evolving, creating uncertainty around deployment timelines,' said Arjun Malhotra, general partner at early-stage venture firm Good Capital.The fintech executive cited earlier said there are expectations that the Reserve Bank of India would take note of the increasing deployment of AI in banking and release some broader guidelines and framework that industry feels more confident about investing in these solutions.'AI hallucination is real, while humans mis-sell products as well; what happens if the AI bot gets gamed by the customer or the bot actually makes a mistake and the customer loses money, these are things which will need further industry-wide deliberations,' said Ramanathan of Hyperface.