Pine Labs’ P3P Turns Agentic Payments Into A UPI Compliance Test
Pine Labs’ P3P lets AI agents execute pre-approved UPI payments, but the launch also surfaces unresolved questions on mandates, user authentication, liability, privacy and stablecoin plans.
Autonomous Checkout Moves From Demo To Payment Rail
Pine Labs has introduced Pine Labs Payment Protocol, or P3P, as a framework for AI agents to complete payments after a user gives advance permission.
The design is built around UPI mandates rather than a fresh approval for every transaction, so the user sets a spending boundary first and the agent then acts inside that boundary.
The operating model combines UPI ReservePay, which Pine Labs links to Single Block Multiple Debit, One Time Mandate, the Grantex identity layer and HTTP 402.
In practical terms, a mandate can reserve funds, Grantex checks whether the agent is genuine and records the transaction trail, and the web-payment layer gives software a machine-readable way to request payment.
The important shift is control timing.
The user still approves the mandate, but the later purchase can happen without the user re-entering a PIN at the moment the AI agent acts.
That makes P3P more than a checkout shortcut; it tests how far India’s payments stack can stretch toward event-triggered, software-led commerce.
Early Uses Show The Promise And The Boundary
Pine Labs is not presenting P3P as a theoretical interface.
Gullak, a digital gold savings platform, is the first live use case.
A user can set a condition such as buying Rs 500 of gold if the price falls below Rs 16,000 per gram, approve the mandate once, and let the agent execute the purchase when the condition is met.
Vijay Sales is testing the protocol in a proof of concept for electronics purchases triggered by a target price.
The company says P3P is currently live only on UPI ReservePay, while cards, net banking, wallets and EMI options sit on the roadmap.
Developer documentation also lists stablecoins as a future payment rail.
Those details make the launch commercially relevant for merchants, fintech platforms and AI-agent developers.
P3P would let an agent move from recommendation to transaction, but only if the merchant, user, bank and protocol layer agree on the conditions that define a valid purchase.
Mandates Become The Regulatory Pressure Point
The regulatory question starts with whether a UPI mandate is being used for a purpose it was originally designed to handle.
Mandates are familiar for scheduled or recurring payments such as subscriptions, SIPs and EMIs.
P3P applies the same concept to one-off purchases that depend on a condition, such as a market price or flash-sale availability.
The Reserve Bank of India’s Digital Payments E-Mandate Framework, 2026 requires Additional Factor of Authentication when a mandate is set up.
It also allows recurring transactions up to Rs 15,000 without a fresh Additional Factor of Authentication after the mandate is created, while transactions above Rs 15,000 still require it.
That leaves a practical gap for agent-led purchases.
If an AI agent tries to complete a single high-value payment while the user is absent, the available disclosure does not show how P3P resolves the extra-authentication requirement.
It also leaves open whether NPCI has separately cleared AI-triggered purchases that rely on UPI mandates.
Liability And Data Disclosure Remain Unsettled
Pine Labs positions P3P as user controlled because the user can predefine spending limits, merchants, triggers and validity conditions.
It also says the system can produce a cryptographically verifiable receipt that supports dispute resolution.
That still does not answer who carries the loss if an agent buys the wrong item, pays at the wrong moment or triggers a transaction the user later disputes.
The question becomes sharper for high-value purchases, where the transaction may be hard to reverse after stock, price or settlement conditions change.
Data handling is the other unresolved point.
P3P expands agentic payments from bill-payment experiments into autonomous purchase decisions, which can reveal prompts, transaction records and spending patterns.
The public record still lacks what transaction data remains with Pine Labs, what may be shared with AI providers such as OpenAI or Anthropic, or whether any of those records can be used to train AI models.
The Watchpoint Is Approval Scope
The next test is not whether AI agents can initiate payments.
P3P shows that the technical path is already being assembled around UPI mandates, identity checks and machine-readable payment requests.
The harder question is approval scope.
Banks, NPCI and regulators will need to decide whether a one-time mandate can safely cover autonomous purchasing across changing prices, merchants and triggers.
Pine Labs’ stablecoin roadmap adds a second unresolved layer because India does not recognise cryptocurrencies as legal tender and the RBI has raised financial-stability concerns about private cryptocurrencies.
For India’s fintech market, P3P is therefore both a product launch and a governance test.
If the consent, liability and data rules become clear, agentic payments can move closer to normal commerce.
If they remain vague, merchants may get a new checkout interface before users and banks have a settled answer on who is responsible when the agent acts.
