Artificial intelligence is moving from the margins of experimentation to the core infrastructure of modern economies. Systems built on large models and autonomous agents are increasingly embedded in the logistics networks and financial markets to healthcare diagnostics and national security.

In some domains, the shift is already visible. During recent conflicts such as Russia-Ukraine and the ongoing tensions involving the United States/Israel and Iran, AI tools have reportedly been used to analyze intelligence streams and compress military decision-making cycles. In science, breakthroughs such as AlphaFold have demonstrated how machine learning can accelerate discovery and research in medicine.

Across these domains, the pattern is similar: systems capable of processing vast datasets, identifying patterns and recommending or executing actions are beginning to reshape how complex decisions are made. It is only a matter of time before this capability reaches another critical layer of the digital economy -payments.

For years, AI in payments has largely been confined to familiar functions such as fraud detection, risk scoring and compliance monitoring. But as AI agents become capable of executing multi-step tasks, interacting with APIs and navigating digital ecosystems, payments themselves may become agent-driven. In other words, AI may not simply optimize payment systems. It may begin to participate in them.

In this edition of FinTales, we examine this emerging intersection between AI and payments. We also explore UPI One World’s potential and RBI’s clarification on default loss guarantee.

Let’s dive in.

Main Course 1: A deep dive into agentic payments and what India's AI-driven payments future could look like.

Main Course 2: A piece on how streamlining of UPI One World can open up new opportunities

Dessert: Sweet news about default loss guarantee.

Mints: Quick refreshers on the latest fintech developments. 

Main Course 🍝

Agents of Change: The Leap from Digital to Agentic Payments

Image credits: AI-generated

For the past decade, India’s payments revolution has been about speed. The Unified Payments Interface (UPI) made money move instantly – turning payments into a near-frictionless utility for millions of users and businesses.

The next shift, however, may be far more profound. We will soon see AI systems capable of completing end-to-end commerce tasks, including making payments on behalf of users through UPI. This emerging model – often described as agentic payments – could mark the transition from instant payments to intelligent payments.

Razorpay showcased this future through pilots launched at the AI Impact Summit 2026. It collaborated with the National Payments Corporation of India (NPCI) to illustrate how conversational AI assistants (operating through voice or text) could discover products, compare options and complete transactions using UPI rails. Other participants in the pilot included AI companies such as Anthropic, e-commerce platforms such as BigBasket and agentic AI companies such as Gnani.ai.

These pilots offer an early glimpse of what agentic commerce could look like. Instead of navigating multiple apps, comparing options and manually authorizing every step of a transaction, users may soon rely on AI assistants capable of understanding simple instructions such as: ‘pay my electricity bill if it is below Rs. 3,000’ or ‘buy the cheapest flight to Delhi tomorrow’. The agent searches for options, evaluates prices and completes the transaction – using UPI rails to move funds in real time.

India’s Unique Position

India may be particularly well-placed to lead this transition. It already operates one of the world’s most sophisticated digital public payment infrastructures through UPI, which processes billions of transactions every month.

Crucially, UPI is not just a real-time payment system. It is also a programmable payments network, with features that allow conditional, automated and delegated payments – capabilities that align closely with the requirements of agentic systems.

Two relatively recent features illustrate this potential:

•       UPI Reserve Pay: allows users to reserve funds for a future transaction without immediately debiting their bank account. Funds are blocked and can be automatically triggered for payment once predefined conditions are satisfied. In an agentic commerce environment, this mechanism could allow an AI assistant to execute payments only when criteria such as price thresholds or time conditions are met.
•  UPI Circle: allows a user to authorize another person to initiate payments from their account within defined limits. While originally designed for family and delegated payment use cases, the architecture closely resembles the delegation framework required for AI agents. In principle, a user could grant limited authority to an AI system to initiate payments within guardrails set by the user.

Taken together, these features already provide the core ingredients required for agentic payments: conditional authorization, delegated payment authority and real-time settlement. The latest pilots demonstrate how these capabilities can be combined with AI systems to enable autonomous commerce flows.

What Agentic Payments Could Solve

If deployed responsibly, agentic payments could unlock several benefits.

•       Financial inclusion: AI assistants could become the primary interface for financial transactions. Instead of navigating multiple apps or payment steps, users could simply instruct an AI assistant to perform tasks such as paying bills, booking travel or purchasing goods. For many users – especially the elderly, first-time internet users or those more comfortable with voice interfaces – this could make digital payments significantly easier to use.
•  Greater efficiency in everyday commerce: Many routine transactions still require repetitive manual steps – bill payments, subscription renewals, ticket bookings or routine purchases. Agentic systems could automate these tasks while optimizing for cost, convenience or user preferences. For instance, an AI assistant could automatically pay utility bills below a defined threshold, purchase groceries when prices fall within a certain range or manage recurring subscriptions without repeated manual authorization. Over time, such automation could significantly reduce friction in digital commerce.
• Operational efficiency for businesses: Agentic payments are not limited to consumer use cases. Businesses could deploy AI agents to manage operational payments such as vendor settlements, inventory-linked procurement, subscription services and recurring vendor contracts. For small businesses and startups, this could reduce administrative overhead and improve working-capital efficiency.
• Machine-to-machine commerce: Agentic payments may also enable direct transactions between machines. Electric vehicles paying charging stations, smart appliances paying utility providers or logistics systems executing automated toll payments are examples of how machines could increasingly transact autonomously. In such scenarios, UPI could become the settlement layer for India’s emerging machine economy.

The Regulatory Questions India Will Need to Answer

While the technology may be advancing rapidly, large-scale deployment of agentic payments will raise important regulatory questions.

•      Who is responsible when an AI agent makes a mistake? Traditional payment systems assume a human decision-maker at the point of transaction. Agentic payments introduce a new intermediary: the AI system that interprets instructions and executes payments. If an AI assistant purchases the wrong product, pays the wrong merchant or is manipulated into executing fraudulent transactions, responsibility will need to be clearly allocated among users, AI developers and payment intermediaries.
•  How should consent and authorization be designed? Payment systems today rely on explicit user consent – typically through OTPs, PINs or biometrics. Agentic payments rely instead on delegated authority, where users grant standing permissions to AI agents to act within defined limits. Designing consent architectures that allow users to set boundaries around transaction value, merchant categories or frequency while maintaining convenience will be essential. Existing UPI features such as Reserve Pay and UPI Circle may provide a useful foundation for such systems.
•  How can fraud risks be contained? Autonomous systems may increase the speed and scale at which transactions occur. Payment infrastructure may therefore require stronger monitoring tools, including anomaly detection systems, transaction velocity limits and human verification for high-value payments.

The Road Ahead

India’s digital payments journey has repeatedly demonstrated how public infrastructure can unlock large-scale innovation. UPI transformed how individuals and businesses move money across the economy.

Agentic payments may represent the next layer in that evolution – one where intelligent systems sit on top of payment rails and execute economic decisions on behalf of users.

If implemented carefully – with strong safeguards, clear regulatory frameworks and continued collaboration between industry and policymakers – India could once again set the template for how the world approaches the next frontier of digital payments.

 Unlocking UPI for Global Travelers 

Image credits: AI-generated

UPI has turned a country once dominated by cash into a place where a roadside tea stall and a luxury hotel use the same payment rail. But there was always one category of users who remained outside this system: foreign visitors.

For years, tourists arriving in India found themselves in a strange situation. The country had leapfrogged into QR-based payments everywhere, yet foreigners still had to rely on international cards or cash. Without an Indian bank account or phone number, they could not easily use UPI.

In 2023, the Reserve Bank of India tried to solve this problem.

It allowed prepaid payment instruments (PPIs) linked to UPI to be issued to foreign nationals and non-resident Indians visiting India. The idea was simple. A traveler could load money in Indian rupees into a wallet and pay merchants through UPI just like a local. The product came to be known informally as ‘UPI One World’.

What is UPI One World?

RBI norms allow regulated entities to issue UPI-linked prepaid wallets to foreign nationals and NRIs visiting India. These wallets:

•       are denominated in Indian rupees and funded through foreign cards or forex conversion.
•       require full KYC (passport and visa verification).
•        can be used for merchant payments via UPI QR codes but cannot be used for P2P transfers or cash withdrawals.
•         must maintain balances below Rs. 2 lakhs, with unused funds convertible to foreign currency.

In essence, the wallet acts as a temporary bridge: it converts foreign money into a regulated INR wallet that can access India’s UPI network.

The Problem

In theory, UPI One World was a neat solution. In practice, it never quite took off. The problem was onboarding. Under the current framework, foreign visitors must complete full KYC before using these wallets. Their passport and visa must be physically verified, typically at airport kiosks or designated counters. Only after this verification can they activate the wallet and start paying through UPI. For a traveler used to tapping a card or Apple Pay, this process feels cumbersome. It requires finding a counter, waiting in line, and completing formalities before making the first payment. As a result, UPI for foreigners exists, but very few travelers actually use it.

The Fix

NPCI now appears to be trying to fix this last-mile problem. Recent reports suggest that NPCI has sought RBI’s approval to allow digital onboarding for foreign travelers. If permitted, visitors could complete KYC remotely instead of visiting a physical counter.

The Opportunity

If implemented well, this could unlock a significant opportunity for India’s payments ecosystem. India receives tens of millions of foreign visitors every year. Today, most of them rely on international card networks or currency exchange. But if digital onboarding becomes possible, UPI could become the default way for tourists to pay across India.

For merchants, this would require no additional infrastructure. The same QR code that works for Indian customers would work for foreign visitors as well. For banks and prepaid wallet issuers, it opens up an entirely new product category: short-term travel wallets that combine foreign exchange conversion with UPI payments.

Fintech companies also have an opportunity here. They could build travel-focused payment apps that allow foreigners to complete onboarding before arriving in India. Airlines, travel portals, and visa platforms could integrate UPI wallet activation directly into their journeys. A visitor planning a trip to India could activate a UPI wallet while applying for a visa or booking flights. By the time they land, payments would already be ready to go. Such integrations could turn UPI into something far more powerful than a domestic payment system. It would become a travel payments layer embedded into the global tourism ecosystem.

The opportunity could become even larger if the framework eventually expands to allow peer-to-peer payments in addition to merchant transactions, enabling visitors to split bills, pay guides or transfer money within travel groups.

Conclusion

UPI has already demonstrated how digital public infrastructure can transform domestic payments at scale. Extending that experience to international visitors is the logical next step.

Dessert 🍨

RBI Clears the Air on DLG Provisioning

In February 2026, the RBI amended its NBFC provisioning directions to allow lenders to factor Default Loss Guarantees (DLGs) into their Expected Credit Loss (ECL) provisioning calculations. This move was widely welcomed across the digital lending ecosystem.

DLGs are common in fintech–NBFC partnerships. When fintechs source loans for an NBFC, they sometimes provide first-loss protection – through a cash deposit, guarantee, or similar structure – promising to absorb the first slice of borrower defaults. This ensures the fintech has skin in the game. While the RBI formally permitted such arrangements in digital lending in 2023, confusion emerged in May 2025 when an informal guidance by RBI suggested that NBFCs should ignore DLG protection when calculating ECL provisions.

That position was counterintuitive. ECL is a forward-looking framework where lenders estimate losses based on the borrower’s likelihood of default, the lender’s exposure, and the expected loss if that default occurs. If a fully funded guarantee absorbs part of that loss, the lender’s expected exposure should logically fall. Yet the earlier guidance effectively required NBFCs to provision for losses already contractually covered by DLGs.

The February 2026 amendment resolves this. NBFCs can now net off DLG coverage while computing ECL provisions in digital lending and co-lending arrangements, provided the guarantee forms an integral part of the loan structure. The impact is meaningful: lower provisioning improves capital efficiency and profitability for lenders, making DLG-based partnerships more viable and potentially enabling more cost-efficient credit for borrowers.

Mints 🍃

🪙 Karnataka HC refuses to quash FIR against Jar: The Karnataka High Court has refused to quash an FIR filed against Jar, allowing the investigation to proceed at this stage. The petition had sought the court’s intervention to set aside the FIR. However, the court declined to interfere in the early stage of the proceedings, noting that the investigative process should run its course before any determination on the merits of the allegations is made. The order effectively means that law-enforcement authorities can continue their probe, while Jar retains the option to pursue appropriate legal remedies as the case progresses. 

🛡️ RBI proposes tightening of customer protection norms: RBI has issued draft proposals aimed at curbing mis-selling and strengthening consumer safeguards. The measures propose banning third-party product bundling and dark-pattern marketing practices, while requiring explicit customer consent for financial products. In parallel, RBI has released draft loan-recovery guidelines to tighten conduct standards for recovery agents and better protect borrowers. Separately, it has also proposed rules requiring banks to compensate customers for digital payment fraud, with payouts capped at Rs. 25,000.

🔎 RBI directs NPCI to probe autopay complaints: The RBI has asked the NPCI to examine rising complaints about erroneous deductions and cancellation issues in UPI Autopay. Users have reported involuntarily created mandates, unclear consent processes, and limited visibility of recurring payments. The concerns come amid rapid growth in Autopay usage, driven by the expansion of digital subscriptions.

🔄 India and Malaysia announce measures to enable cheaper cross-border payments and remittances: India and Malaysia have formalized the Malaysia-Indian Digital Council and announced plans to link India’s NPCI payments network with Malaysia’s PayNet payments network, to enable more efficient cross-border payments and remittances for tourists, students and diaspora-linked remittances.

🌏 Apple readies India payments entry: Apple Inc. is in talks with ICICI Bank, HDFC Bank, and Axis Bank to launch Apple Pay in India by mid-2026. The service is expected to support UPI and card payments. The plan reflects Apple’s push to expand in India’s fast growing digital payments market and compete with players such as Google Pay and PhonePe.

🌐 India expands UPI to Israel: India has extended the rollout of UPI to Israel, marking a significant step in digital finance cooperation. The announcement was made during Prime Minister Narendra Modi’s state visit to Israel, where he met Prime Minister Benjamin Netanyahu to discuss strategic and technology partnerships.

 For any queries, reach out to us at contact@ikigailaw.com