PhonePe expands AI across consumer, merchant and corporate operations

MUMBAI: PhonePe is putting AI to work, but the fintech does not want to sprinkle it everywhere just for the sake of saying it did.

The company is expanding artificial intelligence across its consumer platform, merchant services, software development and corporate functions, with Rahul Chari, Founder and Chief Product & Technology Officer, saying the focus is on use cases where AI can deliver measurable improvements rather than turning every service into an AI product.

Speaking at the Global Fintech Fest 2026, Chari said PhonePe is deploying different AI capabilities across the organisation instead of relying on a single technology or solution.

For consumers, the company is exploring an AI-powered universal search that can understand natural and vernacular language. Potential applications include gas bookings and bill payments in languages such as Hindi, English and Tamil.

PhonePe is also looking at conversational AI for personal finance, allowing users to ask questions about their spending, while similar capabilities are being explored for insurance and investment services. Agentic features are also being considered for products including gold, silver, systematic investment plans, fixed deposits, recurring deposits and pension plans.

The AI push extends to merchants and direct-to-consumer businesses. PhonePe is exploring tools that could simplify website creation, payment gateway integration and connections with platforms such as Instagram and WhatsApp. Its Smartpages product already allows businesses to create websites with payment integration.

The PhonePe for Business app could also get agentic functions for tasks such as answering business and customer queries and creating and distributing offers and coupons.

Behind the scenes, AI is becoming part of PhonePe’s software development pipeline, covering solutioning, coding, code reviews, testing, deployment, troubleshooting and debugging.

The company has built several internal tools for this work. TraceMine is used for crash analysis, Lens brings together logs and deployment information for initial root-cause analysis, while KadAI provides centralised code orchestration and agentic development.

PhonePe has also created an Agent Hub for non-technology teams to build AI-enabled workflows. Designed for a regulated financial services environment, the platform includes safeguards around data access, data leakage and localisation.

It uses an in-house large language model router, open-source models hosted on premises, as well as internal tools and data. Legal, compliance, finance and human resources teams are already using the system.

One of the more ambitious targets is to compress PhonePe’s new product introduction process. Reviews covering compliance, data privacy, legal, trust and safety currently take about 30 days; the company is exploring whether AI can bring that timeline down to potentially less than a day.

Customer support and fraud investigations are already showing more tangible results. PhonePe said automation now handles about 94 per cent of customer support, up from around 90 per cent, with the company targeting 96 per cent.

Around 80 per cent of fraud investigations can also be automated, allowing human teams to focus on complicated cases and emerging fraud patterns rather than routine investigations.

The scale of the operation makes those efficiencies significant. PhonePe processes more than 10 billion transactions a month, with the figure now closer to 12 billion, and operates with about 500 customer support agents.

For Chari, however, the AI scoreboard is not about how many features carry an AI label. The test is whether a deployment produces a tangible improvement in an operating metric, product or process.

In other words, PhonePe wants AI to do more than look intelligent, it wants the numbers to prove it.