PhonePe is redesigning its products and infrastructure around artificial intelligence, focusing on natural language understanding and privacy to redefine digital payment experiences and streamline merchant onboarding.
PhonePe is betting that the next stage of digital payments will be shaped less by screens and more by software that can understand what users mean. At a recent Economic Times discussion at the company’s Bengaluru office, senior leaders said the firm is redesigning products, infrastructure and internal workflows around AI, while keeping privacy and accountability at the centre.
Gautham Krishnamurthy, head of product for consumer platform, said the shift is moving apps away from forcing users to learn menus and jargon, and towards systems that can interpret natural language and act on intent. He argued that this may be one of the biggest changes in app design since touchscreens. In practical terms, that means users could increasingly ask for what they want in their own words, whether that is sending money, checking a failed payment or finding a bill.
PhonePe says it is already seeing more natural-language behaviour in search, helped by AI-powered search tools that were launched to support multilingual use at scale. The company has also said in a separate announcement that the feature was built with Microsoft Foundry and is designed to route users to tasks more directly while keeping data inside PhonePe’s environment. That approach matters in India, where users often mix languages, spellings and local references when searching for services.
The same thinking is extending beyond consumer search. PhonePe has also introduced an AI-powered integration layer for merchants that can cut payment gateway setup from a technical project into a conversational workflow, according to its own release. A separate report by PYMNTS said the tool can reduce integration time from weeks to minutes, a potentially significant gain for small businesses that lack dedicated engineering teams.
Behind the products, the company says it spent years building its AI stack before exposing features to consumers. Santanu Sinha, chief architect, said much of the groundwork involved infrastructure, governance and machine-learning systems that were put in place well before recent AI roll-outs. He said the same platform now supports large language models, recommendation systems, fraud detection and models that can run on users’ devices.
Privacy has been a major design principle, PhonePe said. Sinha said the company deliberately built much of the system in-house because it handles sensitive customer data. Its architecture also allows some functions to run on-device, which reduces the need to move information to central servers. PhonePe has separately described its Guardrails initiative as a way to strengthen risk detection, alert customers and reduce exposure to fraud and malware.
Radhakrishna Ramaseshu, head of engineering for consumer apps, said AI is also changing how software is built inside the company. Developers are using it to draft code, create tests and reduce time spent on repetitive setup work. But he said human review remains final, with compliance and security checks still applying to every workflow. “AI might generate code, build code, but the buck stops at the human,” he said.
That balance also shapes Agent Hub, an internal project that began as a way to help smaller teams search through large stores of company knowledge. PhonePe says the system has evolved into AI agents that can retrieve information, perform actions and work across tools such as Slack and Gmail. The company has also said it is experimenting with ChatGPT through a strategic collaboration with OpenAI to bring more advanced AI features to Indian users across its ecosystem.
The leaders framed AI not as a replacement for the company’s core business, but as a layer that makes it easier to use. PhonePe said it is applying the technology to credit education, fraud detection, developer productivity and customer support, but still sees itself as a product-led company. As Sinha put it, the firm does not intend to become “AI-first”. The aim, instead, is to use AI to make money movement, account management and decision-making simpler without weakening trust.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
