Fintechs race to deploy AI agents, but revenue payoff remains elusive

Paytm, Pine Labs, Razorpay, Cashfree and Juspay are deploying AI agents across merchant and internal workflows, but monetisation remains nascent as most offerings are still being tested

Artificial intelligence (AI) is becoming the equivalent of table stakes for fintech startups in India as companies look to deploy agentic interfaces for their merchants, allow agents to accelerate internal workflows, and eventually improve margins.

 

A wide range of Indian fintech startups, including consumer and merchant-facing companies, are expanding their agentic portfolio as they look to sell more services around their core business.

 

In June, payments major Pine Labs rolled out an agentic payment protocol P3P for autonomous transactions on select merchants. Paytm, one of India's largest digital payments firm, is working on agentic solutions for its merchants.

 

Fintech companies such as Cashfree Payments, Juspay, and Razorpay have rolled out agentic interfaces for their merchants, who originally signed up for digital payments processing and routing, with a pitch to improve success rates for them.

 

AI interfaces sitting on the merchant's backend are now able to automate tasks such as following up on abandoned carts and retrying failed payments, acting on a business owner's plain-language instruction to issue a refund or send a payment link, and make product catalogues discoverable inside customer-facing AI interfaces such as ChatGPT.

Visible impact

Beyond product launches, companies are already seeing the impact of AI adoption on their businesses, with the potential for higher margins.

 

“We do think that AI structurally not only accelerates operating leverage, it also expands the opportunity for higher margins over time, because you're just able to do more with less and we have been able to deploy a huge number of applications and agents within our business which are helping us become very efficient,” said Madhur Deora, president and group chief financial officer (CFO) of Paytm, in a call with analysts.

 

For routine processes such as call centres and customer follow-ups, AI agents are now beginning to deliver results.

 

The growing adoption of frontier AI models is also improving productivity, with a greater share of code now being written with AI.

 

“At least four divisions of Pine Labs, 90 per cent of all new code which is being written using AI,” Amrish Rau, chief executive officer (CEO), Pine Labs, said on an analyst call after the firm’s Q1FY26 earnings.

 

He added that the company was seeing “great demand in the Indian market, especially on the agentic payment side.”

 

“But the flow of that is actually being seen in our online business. Our online business, which is very closely associated with what is happening on the agentic front, is getting a lot of traction in the market,” he said.

 

In July, Business Standard reported that the National Payments Corporation of India (NPCI) was building a new standard for agentic payments on the country’s real-time payments rails Unified Payments Interface (UPI).

 

The new protocol for the agentic world is expected to be called Unified Agentic Protocol (UAP) and is being designed to create a trusted, common, interoperable infrast­ructure through which AI agents can be registered, verified, and authorised to transact across the UPI ecosystem without changing the underlying rails of the payments system.

Where is the revenue?

Fintech founders said that despite a long pipeline of AI launches, most remained in beta or testing phases. This means that companies are yet to see meaningful contributions coming from AI alone.

 

Companies are also evaluating an additional cost squeeze, with the rising cost of servers and technology potentially affecting broader ambitions around AI-driven revenue.

 

“What is happening in the industry is, because of the demand on memory and so on, the cost of servers is actually going up. So now, how that will translate into cloud costs and so on, is yet to play out, because there is a lag. Many of them would have procured servers. The retail cost has suddenly gone up. This will play out over time,” a technology chief at a large payments company said.

 

To level the playing field in the interim, companies are therefore taking a cautious approach to pricing.

 

Many are exploring outcome-linked pricing, where charges are tied to metrics such as the number of successful cart conversions, disputes resolved favourably, or collections completed through the interface.

 

That said, AI is still in its early stages in India, with more companies expected to roll out AI-based solutions to build additional revenue streams beyond their core businesses. 

Miles to go 

• For routine processes such as call centres, AI agents are beginning to deliver results
• Fintech founders said despite a long pipeline of AI launches, most remained in beta or testing phases
• Firms are evaluating an additional cost squeeze with the rising cost of servers 
• They are taking a cautious approach to pricing