Fintech major Razorpay is offering businesses AI agents that can recover abandoned carts as well as respond to payment disputes, with the launch of what it calls the world’s first AI-native Agent Studio for payments at its flagship event FTX26 in Bengaluru on Thursday. These agents have been built for specific tasks, using Anthropic’s Claude Agent SDK, according to the company. In an interview with Anees Hussain, Co-founder and Managing Director Shashank Kumar speaks about the company’s AI-first pivot, why small and medium-sized businesses (SMBs) will be the primary target, the global expansion road map, and where things stand on a potential IPO. Excerpts:
Tell us about the new launches.
Agent Studio is a marketplace of AI agents that businesses deploy in one click — each built for a specific task. We’ve launched agents that recover abandoned carts, respond to payment disputes, retry failed subscriptions, and forecast cashflow. There’s also a ‘Build Your Agent’ feature where businesses create agents in plain English, without code. The Agentic Experience Platform makes onboarding, integration, and operations conversational. Onboarding that earlier took 30-45 minutes now takes just five. Integration happens in under 10. A merchant can upload a bank statement and ask the system to match it with their settlements — what took hours now happens in seconds.
What does token usage look like for these tasks today?
We are willing to absorb the costs of token usage in the beginning. The behaviour shift has to happen first. People don’t even know these products exist. Our pricing will be based on task complexity, but 12 months from now the cost of underlying engines will come down. Similar to how telecom costs fell even as speed increased. The real focus is building user behaviour — when a payment failure happens, someone has to look at a dashboard and act. Earlier, you needed engineers or operations people. AI can take the lead and operate faster.
When an agentic system errs, who’s accountable?
The liability will have to be on us who provide the service. In 10 years, payment failures have been brought under control. Same with AI. There will be errors initially. As companies learn, it’ll come down.
AI model companies are struggling with profitability. Does that worry you?
If a model can’t make money, it will be bought out by someone who runs it better. The genie is out of the bottle. Players may shuffle, but AI is like electricity. Not going away.
Who are these products for — enterprises or SMBs?
For us, enterprises contribute more volumes but lesser margin. SMBs contribute 20-30% of volumes but a stronger revenue line. But the revenue split is close to 50:50. Enterprises want 10 different things and might build in-house. Zomato and others build a lot themselves. SMBs want efficiency and lesser burn. With agentic products, the focus will be on SMBs, startups, and mid-market.
Beyond payments, where is growth coming from?
Payments still accounts for majority of our revenues. Beyond this, we’ve become the third-largest in gift card issuance and distribution in the country. I feel we’ll be second-largest this year. We launched company registration last year — over a thousand registrations a month, with community, mentorship, and networks at nominal cost.
How has global expansion been?
We are doing really well in Malaysia. In Singapore, we hit 50 customers recently. This year we’ll grow these further. These markets have a high spending capacity and ticket size. Beyond that, Southeast Asian markets are a target — the Philippines, Thailand, Vietnam.
What is AI’s impact on internal cost structures?
AI helps reduce the cost of serving customers. If we scale up from 100,000 to 200,000 customers and costs rise in sync, that’s not where we want to go. We’d take the savings and build more products.
Where do things stand on IPO?
We are definitely looking at an IPO. But it’s difficult to put a marker on timeline. With global uncertainty due to the US-Israel-Iran war, we don’t really know. It’s hard to give a status. Those calls are best left to the board. However, if the IPO route doesn’t become viable immediately — we’re well-capitalised and have healthy balance sheet. We don’t need to raise external funds.
In terms of the general perception with regards to the war, in our category businesses aren’t worried. But there is a general concern among our customers. If the war goes on, there will be more impact on the economy. People will feel it.