Beyond Robocalls: How Banks, Brands are Tapping AI Chatbots to Build Trust

Over the last few years, customer-facing businesses have successfully leveraged conversational AI-powered chatbots and virtual assistants to handle a large volume of customer queries, reducing operational costs and contact centre expenses. 

However, voice AI is no longer just about doing away with human agents. Today, consumer firmsare using voice AI to reach out to more consumers, maintain retention, and multiply business growth.

Instead of replacing human touch, enterprises are now asking how AI can scale empathy.

Banking on Trust in Bharat

For Jana Small Finance Bank, which serves microfinance customers across India, voice AI has unlocked a structural opportunity.

Giridhar Amrela, Head of AI and Innovation at the bank, explains to AIM on the sidelines of the ConvoZen AI Summit in Bengaluru that its customers are often not tech-savvy. “The notifications, most of them are in English, so it is very difficult to train them and make them learn.”

The breakthrough came through AI-powered pre-approved loan calls. “We started this with AI, and I think the response was phenomenal,” Amrela says, adding that the number of customers taking loans from other institutions has “significantly fallen” after these AI-driven outreach calls.

For a bank, however, scale cannot come at the cost of compliance. Guardrails are non-negotiable.

“So whenever we call the customers, we won't feed the bot with this information,” Amrela says, referring to KYC details, balances, and card data. “There's no chance that the AI agent can know this information only. Let's say there's a loan call, so only the information regarding the loan offer is being provided.”

Beyond data minimisation, the bank has invested heavily in scenario mapping. “We have about 30–35 edge cases,” he explains, built using historical customer and fraud data. 

The strategy reflects a core insight: in regulated industries, conversational AI must be bounded, not free-floating.

Speaking in Dialects, Not Defaults

But conversational AI cannot be restricted to urban centres. An important emerging theme is localisation, not just language translation, but dialect fidelity. Trust begins when customers hear their own language spoken naturally.

The bank deploys region-specific bots. “If I talk to someone in Bihar or Uttar Pradesh, there's a different bot used. And for someone who is in Punjab or Delhi, there's a different bot used,” explains Amrela.

Voice personalisation goes even further. “We have some voice agents that exactly sound like a dude,” he notes. “We have specialised voices… that have slang of, let's say UP, or Delhi, or Punjab. These are all built in so that they can connect well.” 

The outcome? “Giving them a response in their language builds trust,” he says, adding that the initiative has improved Jana Small Finance Bank’s customer satisfaction scores.

Voice AI in D2C

In the D2C space, too, the ambition is similar.

Nilesh Kambli, Senior Vice President of Customer Experience at beauty brand Pilgrim, describes a relentless culture of feedback pervasive in the consumer space. “In our monthly leadership meetings, our first agenda is understanding what the customer's cohort was in the last month.”

Traditionally, this meant 20- to 45-minute conversations with humans. But that model does not scale. The brand then thought of building a WhatsApp bot or a voice call to talk to customers and get the insights.

For product companies, conversational AI becomes a dynamic research layer embedded into operations.

At Souled Store, returns—often the most painful metric in D2C—have become an AI opportunity. 

Ajay Lulla, Senior Vice President of Customer Experience at the brand, explains that instead of passively accepting returns, the company uses voice AI to understand intent and potentially convert it. “The voice AI can actually come back and ask you that if you're returning it for a particular size… We'll actually pitch an exchange to you,” he says.

The result? “You'll not only retain the customer, but you also come back and actually get some money back into the ecosystem.”

However, he acknowledges that metrics are still evolving, but the experimentation mindset signals that companies are iterating towards measurable ROI.

Do Customers Care if it’s a Bot?

According to a ServiceNow study, 80% of Indian consumers now use AI chatbots to check the status of issues and complaints, signalling acceptance among consumers. 

What customers dislike, Kambli says, is “a bot trying to act smart and go into the loops and frustrating people to the limit.” 

Amrela observes that some customers, after realising they are speaking to a bot, become more candid. “Once they start knowing that they're a bot, now they actually open up to the bot… which is something which we didn't expect.”

But escalation remains essential. “Some customers will ask, ‘Who are you?’ … That's when we transfer,” he explains.

The hybrid model—AI-first and human-in-the-loop—is now becoming the business mantra of customer-facing operations.

The Five-Year View

Will 90% of queries be handled by bots in five years?

Lulla remains pragmatic. “As long as it's available, easy, it's giving you the right answer, and it's efficient… I don't think it will make a difference to them,” he says.

Amrela is more emphatic about growth. “Five years from now, it's definitely going to significantly increase the spread,” he adds. Operational and transactional conversations, he suggests, are primed for automation, while nuanced, case-specific issues will still warrant human interaction.

Kambli captures the middle ground. “For 90% of transactions, operational things, definitely people want to talk to an AI… I would hate to talk to an AI if I have a hard time.”

In other words, conversational AI is being reframed—not as a cold replacement for humans, but as a distributed, multilingual, always-on interface between brands and millions of Indians.