NextFin News - Flipkart's fintech arm Super.money is placing its biggest wager yet on artificial-intelligence agents, betting that the next wave of digital payments will be made by software acting on a user's behalf rather than by a person tapping a screen. The bet lands at the moment India's payments infrastructure is being rewired to let AI agents pay bills, buy groceries and execute trades inside pre-set limits without asking for permission on every transaction.
The stakes are defined by a stark imbalance. Super.money is the fifth-largest app on the Unified Payments Interface, with more than 15 million customers, while Google Pay and Walmart's PhonePe together account for around three-fourths of UPI's monthly transaction volume. PhonePe alone holds roughly 46% of the market. The upstart's answer is not to outspend the giants on user acquisition. It is to move the battlefield from the payment itself to the decision that precedes it.
The Infrastructure Shift: Payments Without Per-Transaction Approval
India is preparing a framework that would let AI agents make small digital payments without requiring user approval for every transaction, according to people familiar with the matter. The Unified Agent Protocol is expected to be unveiled at the 2026 Global Fintech Fest in Mumbai, and would put India among the first countries with national infrastructure for agentic AI payments.
The National Payments Corporation of India, which operates UPI, is examining how digital agents can be identified and authorised inside the payments ecosystem. "NPCI is also examining the protocols that may be required to identify and authorise digital agents within the UPI ecosystem while preserving interoperability, auditability and settlement finality," Ajay Kumar Choudhary, NPCI's non-executive chairman, said at the Global Fintech Fest 2026.
The framework is expected to build on existing UPI capabilities: UPI Circle, which lets a primary account holder delegate payment authority, and Reserve Pay, which enables funds to be blocked for multiple debits. Customers would set rule-based instructions for agents on when and how much to pay, with spending limits, identity checks and audit trails built into the architecture. A liability mechanism is also needed as responsibility shifts between users, agents, banks and merchants.
"An agent may interpret intent, but authoriser must verify identity, must verify mandate, must verify limits, and consent," Choudhary said. The architecture should therefore separate intent, authorisation and settlement.
The scale of the network makes the shift consequential. UPI processed 24.51 billion transactions worth Rs 29.82 lakh crore ($314.21 billion) in August 2026 alone, according to data from the payments operator. Low-value, frequent purchases such as groceries are likely to be among the first use cases, with more advanced applications - agents making purchases based on sale prices or executing investments when specified price thresholds are met - coming later.
Super.money's Wedge: Commerce, Credit and Payments in One Loop
Super.money's AI-agent bet is not a standalone product announcement. It is the top layer of a stack the company has been assembling deliberately, and the economics of that stack explain why a smaller player would pick this fight.
The payments app is the acquisition hook, not the profit centre. Around 80% of Super.money's revenue already comes from personal loans, about 10% from credit cards and the remaining 10% from payment products such as bill payments and recharges. The fintech says it retains roughly 85% of users, with 60% to 70% of its transactions coming from customers under 30.
In August, Super.money launched splitStore, an in-app marketplace that lets customers buy products across smartphones, fashion, electronics, home appliances and furniture using zero-interest instalments. The marketplace features around 6 million products from brands including Apple, Nothing, Nike, Adidas, Marshall, Mokobara and Snitch, and leverages Flipkart's delivery network for fulfilment. The company expects splitStore to contribute 20% of total business by December.
The margin arithmetic is the point. Unlike checkout finance, where a company's margin is likely to be around 2% to 3%, the splitStore model can generate a margin of 10% to 12%, founder and chief executive Prakash Sikaria said in an interview.
"We are becoming an affiliate partner to Flipkart or D2C brands. We don't want to get into buying or selling. When a customer buys a product, at the back-end it is fulfilled and delivered by Flipkart or the brand. This is more like Google Shopping, where the feed is coming from everyone," Sikaria said.
The credit engine is being scaled in parallel. Super.money has teamed with Kotak811 to issue secured credit cards, aiming for about 2 million cards in the next 12 months - roughly 60% to first-time borrowers - and 5 million within two years, at a pace of about 200,000 a month. Sikaria expects the Kotak alliance to contribute around 10% of revenue next year as the company works toward profitability by 2026. The company is targeting $100 million in annual revenue by 2026, having been on track to close 2025 with around $30 million in annual recurring revenue.
There is also a checkout layer. A partnership with SoftBank-backed Juspay powers Super.money Breeze, a one-click checkout experience for online merchants aimed primarily at direct-to-consumer brands, with about 1,000 merchants already using the solution.
Why the Agent Layer Changes Who Captures the Value
The first-order reading of agentic payments is simple: consumers will save time because software will handle routine purchases. The second-order consequence is who stands between the consumer's intent and the merchant's checkout.
Today, a UPI app is a pipe. It executes an instruction the user has already formed. In an agentic model, the agent forms the instruction - it evaluates options, applies the user's rules, selects the discount and completes the transaction. The agent therefore sees demand before the merchant does, and it controls the payment rail and, in Super.money's design, the credit decision at the moment of purchase.
That is why payment volume is best understood as an acquisition cost rather than a revenue line for Super.money. The monetisation sits in the credit spread on personal loans - already 80% of revenue - and in the 10% to 12% commerce margin on splitStore. UPI's near-zero merchant discount rate makes a pure-payments strategy a volume game the giants are built to win. A credit-and-commerce stack is a margin game where a focused player can compete on a smaller user base.
Sikaria has been explicit about the targeting. Super.money is deliberately focusing on India's top 10 million to 30 million users, rather than competing with mass-market payment players such as Google Pay or PhonePe that target hundreds of millions.
"Our focus is to bring in users who have a higher propensity to engage with our products," Sikaria said. "UPI happens to be the core engagement and acquisition hook, but for people who don't want to engage in financial services or other products that we launched, we do not want to serve them from a UPI or payment perspective."
The affiliate model is also a defensive fence. Because splitStore does not buy or hold inventory - Flipkart or the D2C brand fulfils every order - Super.money avoids the balance-sheet risk of e-commerce while still capturing the commerce margin. If the agent layer eventually aggregates demand across merchants and shifts bargaining power toward the platform, the company is positioned to capture that value without taking inventory risk.
Cyclical Wave, Structural Shift: Separating the Two
The agentic-payments shift is structural, and Super.money's current traction is riding a cyclical credit wave. Confusing the two produces the wrong conclusion.
The structural case rests on three pieces of evidence. First, this is an infrastructure change: the Unified Agent Protocol is a permanent new rail being built into UPI, not a promotional feature. Second, it is a behavioural regime change: delegating spending authority to software alters the consumer-firm relationship in a way that does not self-correct. Third, there is a precedent: UPI itself displaced cash and cards for small-ticket payments, and UPI AutoPay displaced manual bill payment. Once agents can act on objectives rather than explicit taps, the value migrates from the interface layer - the app screen - to the agent, the rail and the credit stack that executes the objective.
The cyclical leg is separate. Super.money's growth has been powered by a credit-expansion wave - secured cards and buy-now-pay-later demand among young, first-time borrowers. That wave can mean-revert if delinquencies rise or if the Reserve Bank of India tightens digital-lending rules, as it has done before. Which company captures agentic commerce in this cycle depends on credit quality and capital availability, even if the direction of travel is fixed.
Capital is a live constraint. Flipkart has invested about $50 million in Super.money to kick off operations, including a $30 million injection in September 2025. The company is keeping monthly burn at a "low single-digit million number," Sikaria said, and plans to raise additional capital - possibly from external investors. "We need more capital for at least a couple of years," he said. "Very soon, we'll start formulating our capital-raise strategy."
The Strongest Case Against the Leapfrog Thesis
The counter-thesis is straightforward and well-founded: scale decides platform wars, not technology. PhonePe holds around 46% of the UPI market, and together with Google Pay it processes roughly three-fourths of monthly UPI volume. Both giants have deeper pockets, distribution reaching hundreds of millions of users, and established credit and wealth products. They can adopt the same NPCI protocol on day one and bundle agentic shopping into their dominant payments apps. Super.money's deliberate focus on the top 10 million to 30 million users caps its addressable market and leaves it exposed if a giant decides to compete for the same credit-worthy urban consumer.
The answer is that Super.money is not trying to win the mass UPI war. It is trying to win the full financial stack of the credit-worthy urban consumer, where margins live. The secured-card franchise - scaling to 200,000 cards a month through Kotak - and splitStore's 10% to 12% margins are designed for profitability per user, not transaction volume. And the NPCI protocol is interoperable by design: because the rail is the same for every app, the switching cost for a consumer to let a smaller app's agent act on their behalf is lower than it would be in a closed system.
There is a specific signal that would prove the counter-thesis right. Six months after the Unified Agent Protocol launches, if PhonePe and Google Pay together account for more than 70% of agentic-payment volume - roughly their current share of UPI - and Super.money's splitStore contributes less than 10% of total business by December 2026, well below the company's own 20% target, the leapfrog thesis fails. The giants will have absorbed the new rail, and Super.money's commerce-and-credit wedge will not have differentiated.
What to Watch: Three Horizons
In the short term, the proof points are internal. Investors should watch splitStore's contribution to revenue against the 20% target for December 2026, and the pace of secured-card issuance against the 2 million-card goal for the next 12 months. These numbers test whether the commerce-plus-credit stack actually converts users into profitable relationships.
Over the medium term, the race is defined by the protocol rollout. If NPCI delivers the Unified Agent Protocol at the Global Fintech Fest 2026 with merchant integration infrastructure, spending limits and a liability framework, the competition shifts from "who has the most users" to "whose agent is trusted to spend." NPCI's own stated challenge is ensuring that the flexibility of AI does not undermine the certainty required in payments - and the liability mechanism will determine how fast merchants and banks adopt.
Structurally, the company that owns the consumer's spending agent owns the top of the commerce funnel. The likely beneficiaries are platforms that combine payments, credit and commerce in one loop - Super.money's stated design, alongside PhonePe, Google Pay and Amazon Pay. The exposed are pure-play payment apps with no credit book, and merchants who risk becoming price-takers to agent platforms that aggregate their demand.
Three scenarios frame the outcome. In the base case, agentic payments roll out through 2026 and 2027; Super.money carves out a profitable niche among the top 10 million to 30 million urban consumers, reaches its $100 million revenue target, but does not dislodge PhonePe or Google Pay from mass-market UPI. In the upside case, the agent layer resets consumer loyalty, Super.money's credit-first stack captures a disproportionate share of high-value agentic commerce, and the planned external capital raise prices the company at a unicorn valuation. In the downside case, the protocol is delayed or constrained by liability concerns, credit delinquencies among young borrowers rise, and Super.money remains a niche affiliate - leaving Flipkart's roughly $50 million investment with modest returns.
Flipkart's fintech upstart is not trying to beat PhonePe at being PhonePe. It is betting that the next payments war will not be fought over who processes the tap, but over who owns the agent that decides the tap should happen at all.
Explore more exclusive insights at nextfin.ai.
