Strategy memo · Internal · April 2026

India commerce and
agent payments.

India sets the rules. Pine Labs ships the market.
A six-month plan and a twelve-month roadmap.
FromTanya and Umang
ForPine Labs leadership
DateApril 2026
Length8 slides · 7 appendix
1.05Mmerchants on Pine Labs today
1.93Mcheckout points across India
₹11.4Tprocessed annually
716consumer brands on Setu & Plural
Nov '25NSE listed · public company
"Agentic commerce suffers from having been overhyped too early in some corners."
— Stripe 2025 annual letter, Feb 24 2026
·India & Agent Payments 02 / 08 · The map

The rules of the land.

Every protocol that matters, plotted by what is actually possible in India and what is commerce vs. payments. Dark tiles are live with Pine Labs. Faded tiles we watch but do not build.
What Pine Labs ships next→ Slide 7 roadmap
Column 01Possible in India now
Column 02We bring to India · 2026-27
Column 03Live elsewhere · not India
Row A Shopping protocols How a product moves from catalogue to cart.
Pine Labs MCPLIVE · Aug '25 Razorpay MCPLIVE · Apr '25 Cashfree HereLIVE · Feb '26
ACPStripe + OpenAI · Launched Sep '25 UCPLive globally Jan '26 · IN via Flipkart H2 '26
Stripe Commerce ProtocolUnverified · treat as ACP x402Coinbase · crypto rails
Row B Payment protocols How money actually moves and who holds the consent.
UPI AutoPayNPCI · Live UPI CircleNPCI · Live Cards + OTPRBI · Live
AP2Google · Launched Sep '25 Visa TAPLaunched Oct '25 Mastercard Agent PayAggregator cohort MPP IndiaLaunched Mar '26 · Appendix E UPI Reserve PayLive · Oct '25 (SBMD)
Visa Intelligent CommerceUS first · India 2027+ Native crypto railsRBI has not approved
Pine Labs live Protocol live, integration needed Monitor only
Sources · NPCI circulars; public launch announcements Oct 2025 through Mar 2026; Pine Labs product records. 02
·India & Agent Payments 03 / 08 · Fitness test

Protocol by protocol. Is it real in India?

One row each. Decisions, not descriptions. Our reading of which protocols actually run on Indian rails inside the window that matters.
Protocol Owner Live where India verdict Fit with our rails Blocker What we do
ACPAgentic Commerce Protocol Stripe + OpenAI US live · Sep '25 Coming H2 '26
Transaction layer scaled back by OpenAI Mar '26. Discovery layer remains strategic. We ship both.
UPI AutoPay, cards. No regulatory conflict. None material.
ShipWe ship the India ACP discovery + checkout endpoint. Q3.
UCPUniversal Commerce Protocol Google + Shopify Global Jan '26 · IN Q3 '26 Live Q3 '26
Shopify India rollout anchors it. Flipkart is one of 20+ endorsing partners.
UPI native. Shopify MCP adapter path. None material.
ShipWe ship UCP to Indian merchants before Shopify India rolls it out.
AP2Agent Payments Protocol Google Wallet Global · H1 '26 Identity layer, portable
Complements ACP and UCP, not either-or.
Bridges agent identity across protocols. None.
ShipWe contribute the India profile of AP2 to Google. We become the reference.
Mastercard Agent PayAggregator cohort Mastercard IN cohort · Feb '26 Live now
Cashfree, Juspay, PayU, Razorpay are in. We are not.
Cards rail, India-ready. Membership gap.
LeadWe lead the next phase of the cohort by July.
Visa TAPTrusted Agent Protocol Visa Global launch Oct '25 Likely 2027
Visa India moves slower than Mastercard.
Cards rail. Works once live. Visa India timeline.
MonitorSandbox integration. Monitor monthly.
MPP IndiaMachine Payments Protocol Tempo Labs + Stripe Launched Mar '26 (Stripe + Tempo) Adaptation only
Base protocol is crypto. Adapt to stablecoin + UPI.
Fit via Tempo bridge. Appendix E. RBI stablecoin stance.
ShipWe ship the India stablecoin bridge reference. First in market.
Cashfree HereAgent checkout Cashfree IN · Feb '26 Competitor live
Proves the category. Different positioning.
Same rails. Not a partner.
MonitorNot an integration. Competitive watch.
Razorpay MCPMerchant MCP server Razorpay IN · Apr '25 Competitor live
Consumer-first. We lead in B2B.
Same protocol stack as ours.
MonitorNot an integration. Appendix D.
Pine Labs MCPOur merchant MCP Pine Labs IN · Aug '25 Live · ours
OpenAI's first Indian B2B payments partner.
UPI, cards, BNPL, EMI native.
ShipExtends to ACP & UCP endpoints this half.
Stripe Commerce ProtocolAnnounced Feb '26 Stripe Unverified Same as ACP
ACP is the Stripe protocol. No separate launch expected.
Announcement only.
x402HTTP payment protocol Coinbase Global · crypto Not India
RBI has not approved crypto rails.
Rails unavailable.
Visa Intelligent CommerceFull stack agent Visa US · pilots India 2027+
Visa has not prioritised India.
Would fit cards rail once live. Timeline.
MonitorWatch quarterly.
Verdicts are ours. Push back where you disagree. Reasoning expanded in Appendix A. 03
·India & Agent Payments 04 / 08 · Three choices

Three ways we can play. Not three products.

Each is a posture. Each has a cost. We recommend A plus C, and we explain why B is a trap.
We do this
Choice A

Agnostic integrator

Support every protocol that matters. Merchants come from anywhere. We are the rail, not the gateway.
Gain
All major card networks now endorse multiple protocols. Agnostic posture is validated by the market. Near-term revenue on every agent that ships. Neutrality keeps all doors open.
Give up
Thinner differentiation at the protocol layer. Compete on execution and reliability.
Cost
4 to 6 protocol handlers in production by October. Chandan's team.
Revenue now · Moat elsewhere
We do not
Choice B

Merchant MCP builder

Build our own merchant-side stack. Catalogue, inventory, logistics, refunds, payments. End-to-end.
Gain
Own the full commerce experience. One throat to choke for merchants.
Give up
Fight Shopify on catalogue and storefront. Shopify is already ahead. We will lose.
Cost
24 to 36 months. Deep product investment far from core. High opportunity cost.
Wrong fight · Skip
We do this
Choice C

Partner and own the moat

Partner with Shopify and equivalents for catalogue. Own the payments moat underneath. Omni-token vault.
Gain
Distribution via Shopify India. Differentiation on payments, where we actually lead.
Give up
We do not own the storefront layer. Accept this.
Cost
Shopify + Plural integration by August. Omni-vault MVP by October.
The moat lives here
Recommendation
A plus C. We integrate every protocol that ships. We build the omni-vault moat underneath. We do not react to whoever wins the agent layer. We become the reason they ship on Indian rails.
Shopify ahead on catalogue and storefront is Tanya's call. We agree. 04
·India & Agent Payments 05 / 08 · Scope

Where we play. Where we don't.

Two columns. Read straight across. Discipline matters here; the bottom of each column is a list of things we explicitly will not do.
Every item in the Play column is in Slide 7's roadmap.

Commerce side

Play
Payments infrastructure for agentic checkout.
Every agent that books a flight, refills a prescription or reorders FMCG routes through rails. We are the rails.
Play
Omni token vault for multi-merchant consent.
One user authorization. Many merchants. See slide 6.
Partner
Catalogue and storefront via Shopify India, via Plural.
Shopify launched India in April. Their commerce stack is ahead. Plural is the checkout they use.
Partner
Logistics and delivery via Shiprocket.
Shiprocket covers 29,000 pincodes. We do not rebuild this.

Payments side

Play
Agentic-ready UPI: AutoPay, Circle, Reserve Pay.
India's core rail. We have the NPCI relationship to co-author what comes next.
Play
Mastercard Agent Pay + Visa TAP integrations.
Cards rails for agent transactions. Aggregator cohort seat by end of July.
Play
MPP India handler (inbound stablecoin→UPI thesis).
Appendix E. Thesis only, revisit October.
Play
Omni token vault (our moat, slide 6).
Passkey-backed. Cross-merchant debit. India-native.
Play
Agent collections layer.
Agent-to-agent payment primitive. The gap Appendix F-1 documents. Six-month build.
Skip list is deliberate. Strategy is what we say no to. 05
·India & Agent Payments 06 / 08 · Moat

The moat: an omni-token vault.

Passkey-backed. SBMD-native. Cross-agent portable. We ship this first. Nobody else in India has it.
The flow
One user · one authorization
↓ passkey + SBMD ↓
Pine Labs Omni VaultConsent, limits, merchant rules
↓ agent-initiated debit ↓
ChatGPT Claude Gemini Perplexity
→ routed to merchant →
Blinkit Quick commerce 1mg E-pharmacy DMart Ready FMCG reorder MakeMyTrip Travel Nykaa Beauty
The user authorizes once. They never re-authenticate for the same agent or the same merchant. Payments happen within their stated rules. The vault enforces them.
01

Passkey-backed consent

RBI Notification 12898 (effective April 2026) allows passkeys as the authentication factor. User proves identity once with their device. Pine Labs holds the signed authorization.

02

Cross-merchant debit without re-auth

Agent X can spend up to ₹5,000 with merchant Y per week. User sets the rule. Vault enforces it. Covered by NPCI's SBMD primitive (branded as UPI Reserve Pay) for UPI, and by Mastercard & Visa token rails for cards.

03

Cross-agent portability

Same vault serves ChatGPT, Claude, Gemini, Perplexity and any future agent that implements ACP or UCP. Users do not rebuild trust for every agent.

04

Why we ship this first

Razorpay is consumer-checkout. Cashfree is session-based. Juspay is orchestration. None are building a cross-merchant agent-consent vault. We are. Our own MCP today is subscription-heavy — 22 of 31 tools. Agent-initiated one-shot commerce requires a different primitive. That primitive is the omni-vault. Agent collections — "agent collects from another agent" — is the one gap nobody else has filled. Appendix F-1.

Technical architecture in Appendix B. This slide is the argument. 06
·India & Agent Payments 07 / 09 · Roadmap

The product roadmap. Built in India. Shipped to the world.

We do not wait for the market. We ship the market. Here is what Pine Labs ships over the next twelve months.
"Pine Labs MCP today: 22 of 31 tools are Subscriptions. Agentic commerce needs different primitives. Here is what we ship."

Ship in 90 days

May → Jul 2026
Aggressive but real · 5 products incl. Collections · dev PG plan on Slide 08
June

Pine Labs Agent Checkout v1

One API. Any Indian merchant agent-ready in under four hours . UPI, cards, BNPL, EMI behind a single endpoint. ACP and UCP compliant.

June

Pine Labs Agent Collections

Agent-to-agent payment rail. Closes the gap no competitor has filled. Merchant agent collects from consumer agent. Pre-funded, consent-gated, ACP-compliant. See Appendix F-1.

July

Pine Labs Vault (sandbox)

Passkey-backed consent. Per-agent per-merchant rule engine. SBMD-native on UPI. Tokenized on Mastercard and Visa. The moat.

July

Agent Identity Bridge

AP2-compatible identity, portable across ChatGPT, Claude, Gemini, Perplexity. Any agent that asks to debit authenticates through Pine Labs.

July

Merchant Accelerator Kit

Catalogue schema, inventory stubs, delivery webhooks. Everything a mid-market Indian merchant needs. Free.

Developer PG plan has its own slide — see Slide 08.

Ship in 12 months

Aug 2026 → Q1 FY28
India-first moonshots · 6 products · no precedent in India
Q3

Pine Labs Agent Wallet

Regulatory build. Per-agent wallet with pre-loaded INR. QuickFin PPI license does not cover per-agent wallets today. Regulatory work with RBI and NPCI required. We are the Indian entity driving this conversation.

Q3

Omni-Vault for Enterprises

Corporate agent payments. Finance sets rules. AI agents pay vendors, reconcile invoices, manage subscriptions. B2B version of the consumer vault.

Q4

Reserve Pay Protocol SDK

NPCI working group seat active. We author the reference implementation and submit for ratification. Every Indian PSP that follows implements our spec.

Q4

Merchant Trust Score

Credit Plus evolves. Real-time trust decisioning on agent-initiated transactions. Agent says "buy," Pine Labs decides in 200ms, transaction completes.

Q4

Cross-Border Agent Payments

Global agent pays Indian merchant. INR settlement. Stablecoin bridge where legal. Inbound travel, diaspora remittance, cross-border SaaS. Appendix E.

Q1 FY28

Agentic Checkout Standard for Bharat

Pine Labs publishes the open standard for agentic commerce on UPI. Co-authored with NPCI. Every PSP that supports agents implements our spec.

Six of these ten products have no precedent in India. We ship them first. The market follows.
Some of these will look crazy in April. They will look inevitable in October. 07
·India & Agent Payments 08 / 09 · Six-month plan

Six months. Three workstreams.

Milestones only. No revenue projections. Each workstream has one owner and one review point.
May · Jun · Jul · Aug · Sep · Oct
Review: end of October 2026
Workstream 01

Protocol integrations

Owner · Chandan & team
MayInternal state assessment. OpenAI sandbox conversation initiated. UG joins 11 May.
JuneACP endpoint in internal development. Mastercard cohort conversation opened.
Current internal priority: medium. This deck proposes elevating to high. UG advocates sales-side, Chandan's team technical lead. Monthly review.
JulACP sandbox live with OpenAI. Mastercard cohort membership secured .
AugUCP endpoint in sandbox. AP2 POC.
SepVisa TAP sandbox conversations initiated.
Oct3 protocol endpoints in sandbox, one production-ready.
Review · 3 in sandbox, 1 production-ready
Workstream 02

Omni token vault, MVP

Owner · Plural + Setu jointly
MayTechnical spec drafting with Plural + Setu.
JunePasskey registration flow designed. NPCI reference architecture v0 submitted.
JulPasskey flow in staging. NPCI reference architecture v1 circulated.
AugRule engine in staging. Cards rail integration begins.
SepCross-merchant debit MVP in staging with 1 pilot merchant.
OctVault MVP in sandbox with one merchant. Path to production clear.
Review · MVP in sandbox, one merchant
Workstream 03

Merchant design partners

Owner · Tanya
MayQualify 5 candidate merchants. First conversations initiated.
June2–3 merchants in active qualification. First MoU drafted.
JulFirst pilot MoU signed (Quick Commerce merchant).
AugIntegration work begins with first merchant. Second MoU signed.
SepFirst merchant in testing. 3 merchants in active pipeline.
Oct1 merchant live in sandbox, 3 in integration, 2 in pipeline.
Review · 1 live in sandbox, 2–3 integrating
October
October state: one merchant live in sandbox, three in integration, Mastercard cohort active, three protocols shipped, NPCI reference architecture ratified. From Q4, we ship the twelve-month roadmap on Slide 7.
Re-allocated capacity. Not net-new hiring. Full partner tracker maintained by Umang, reviewed weekly. 08
·India & Agent Payments 09 / 09 · The ask

Three asks. One page.

If we agree today, we start Monday. Named owners, one review point, real merchants in the room within thirty days.

01 · Align

The posture

Pine Labs leads India agentic commerce. We ship the protocol. We ship the vault. We ship the roadmap on Slide 7.

We set the pace, not react to it.

02 · Commit

Ship the 90-day products

Pine Labs Agent Checkout v1. Vault sandbox. Agent Identity Bridge. Merchant Accelerator Kit.

Re-allocated capacity. No net-new hiring until we have merchants live.

03 · Open

Merchant rooms and protocol rooms

Tanya opens five merchant conversations. Umang opens NPCI, Mastercard, Google, OpenAI, Anthropic.

We go to market with a roadmap, not a request.

This is what we ship. Ambitious enough to matter. Realistic enough to deliver. By October, we will know which parts were right and which parts were crazy. Push back where we're wrong.
Pine Labs internal · April 2026 · Tanya & Umang 09
Appendix · A to F-2

Open as
needed.

Each appendix answers one likely question. Page numbers match. If a reader lands in the appendix directly, the cross-references back to the main slides are in each header.
A · India protocol fitness reasoning D · Competitor positioning B · Omni vault technical architecture E · MPP India adaptation thesis C · Merchant readiness criteria F-1 · Agent collections gap F-2 · Tri-rail agent payment architecture
Appendix A·Per-protocol India reasoning A / F-2 · Reference

The reasoning behind slide 3's verdicts.

Four regulatory hinges. If these shift, the map changes. We watch them monthly.

RBI Notification 12898

Effective April 2026

Passkeys permitted as authentication factor. Removes OTP friction for authenticated-agent transactions under pre-set limits. This is the unlock for the omni-token vault.

Live · foundational

UPI Reserve Pay

Live · NPCI Circular 228 (Oct 2025)

Reserve-and-debit pattern for delegated transactions. Built on the SBMD primitive (Single Block Multiple Debits). User blocks an amount for future agent-initiated debits without re-authentication. Foundational for agentic UPI.

Live · foundational

India AI Impact Summit

Feb 2026

16–21 Feb 2026 at Bharat Mandapam, New Delhi. Mastercard completed the first authenticated agentic commerce transaction in India at the summit. Aggregators on stage: Cashfree, Juspay, PayU, Razorpay. Issuers: Axis Bank, RBL Bank. Merchants: Swiggy, Instamart, Vi, Tira, Zepto. Pine Labs was not included. We close this gap in 6 to 8 weeks.

Membership gap

RBI stablecoin stance

No official position

RBI has not approved native crypto rails. Stablecoin bridges to UPI may be tolerated but are not endorsed. MPP India thesis (Appendix E) depends on this.

Watch · quarterly
Sources · NPCI Operating Circulars; RBI Notifications; Mastercard India newsroom Feb 2026; MeitY event coverage Feb 2026. A
Appendix B·Omni token vault architecture B / F-2 · Reference

The vault, end to end.

Two boxes. Left, the authorization layer. Right, the debit layer. Each listed in the order traffic flows.

01 · Authorization

IdentityPasskey on user device, registered at Plural checkout. Follows RBI 12898.
ConsentSigned authorization: per-merchant rule, per-agent rule, per-period cap. Held at Pine Labs.
Agent identityAP2 (Agent Payments Protocol) layer. Built on A2A (Agent2Agent Protocol, Google-led). A2A handles task delegation between agents; AP2 handles payment authorization. Pine Labs supports both. Uses three cryptographically signed mandates: Intent Mandate (user's original instruction with rules), Cart Mandate (specific transaction approval), Payment Mandate (settlement rail). Provides tamper-proof audit trail.
Rule engineEvaluates incoming debit against the consent. Sub-100ms. Returns signed approval or denial.
Audit trailEvery decision logged. User can revoke or tighten any rule from the Pine Labs app.

02 · Debit layer

UPI railSBMD primitive (Single Block Multiple Debits, branded UPI Reserve Pay). User blocks an amount for future debits to a specific merchant. Agent debits within the block without re-authentication. NPCI Operating Circular 200 (Jul 2024), enhanced as UPI Reserve Pay via Circular 228 (Oct 2025).
Cards railMastercard & Visa network tokens. Works with Agent Pay and TAP natively.
BNPL / EMIPre-approved credit line. Agent request evaluated against the credit rules, not re-underwritten.
Merchant settlementSame-day settlement to merchant account. Standard Pine Labs flow.
Dispute & refundMerchant refund API returns to vault balance. Zero change to merchant-side flow.
MVP spec reviewed weekly with Plural and Setu tech leads. Full internal doc by end of May. B
Appendix C·Merchant readiness criteria C / F-2 · Reference

What it takes for a merchant to work with an agent.

Five required capabilities, two nice-to-have. Candidate sectors ranked by fit.
01Catalogue schemaStructured product data with SKUs, variants, images, structured attributes.Req
02Real-time inventoryLive stock signal per SKU per pincode. Not refreshed hourly.Req
03Payment rail APIAccepts Pine Labs token. Handles UPI, cards, BNPL uniformly.Req
04Delivery trackingOrder-to-doorstep event stream. Agents surface this to users.Req
05Cancel & refund APIProgrammatic. Agent can cancel on behalf of user within policy.Req
06Saved-address serviceMulti-address, agent-readable. Speeds time-to-cart.Nice
07Loyalty / wallet APICredit balance accessible to the vault, when eligible.Nice

Candidate sectors · ranked

Quick commerceBlinkit · Zepto · Instamart7/7
E-pharmacy1mg · PharmEasy · Apollo 24/76/7
FMCG reorderDMart Ready · BBNow · JioMart5/7
TravelMakeMyTrip · Ixigo · Cleartrip5/7
BeautyNykaa · Purplle · Tira5/7
Tanya owns qualification conversations. First five live by end of October. C
Appendix D·Competitor positioning D / F-2 · Reference

Three different positions. We do not duplicate.

Each has a distinct strength. We keep what is ours; we do not chase theirs.
Company Lead position Agent strategy so far Weakness Our posture
RazorpayPublic · consumer-led
Consumer checkout. Largest agent-ready merchant base on day one. MCP shipped Apr '25. NPCI+OpenAI+BigBasket pilot Oct '25 (still private beta). Claude+Zomato+Swiggy+Zepto Feb '26. Inside Mastercard cohort. Replit India payments partner Feb '26 — first vibe-coding PG in India. Thin on B2B, low developer trust outside checkout. Do not chase consumer. Compete on B2B and developer tooling.
CashfreePrivate · PSP
PSP-first. Agent checkout product (Here) live early. Agentic Payments launched at GFF Oct '25. Cashfree Here launched 17 Feb 2026 at India AI Impact Summit with Mastercard + Swiggy. Live on ChatGPT + Claude. Inside Mastercard cohort. Smaller merchant footprint. No B2B depth. Partner conversations open. Not a competitor on B2B.
JuspayPrivate · orchestration
Orchestration layer for enterprise checkout. Co-author with NPCI. Inside Mastercard cohort. Not building merchant-facing agent tooling. Interop discussions. They orchestrate; we are the vault.
PaytmPublic · consumer-led
Consumer super-app with merchant + fintech stack. Perplexity India partnership (Feb 2025). Not yet an agentic product. No agent-native rails shipped. Cohort seat unclear. Monitor quarterly. No overlap on B2B.
Pine LabsPublic · us
B2B leader. 1.05M merchants. Developer stack (Setu, Plural). Public since Nov 2025 IPO. OpenAI's first Indian B2B partner (Razorpay was first Indian payments partner Oct '25). Two MCP servers live since Aug '25. API Execution server (31 tools: 22 Subscriptions, 5 Payment Links, 3 Orders, 1 Checkout, 1 UPI). Code Generation server (20+ tools). Supported on Claude Desktop and VS Code. Not yet on Cursor, Lovable, Bolt, v0. Setu agentic bill payments live on Claude + ChatGPT since Dec '25. OpenAI B2B partnership Feb '26. ACP & UCP handlers in plan. Omni-token vault as moat. Not yet in Mastercard cohort. Closing in 6–8 weeks. Lead B2B. Integrate everywhere. Own the vault.
Walmart + OpenAIGlobal cautionary data Instant Checkout pilot inside ChatGPT, Oct 2025. 3× lower conversion vs walmart.com direct (Wired, Mar 2026). ~12 Shopify merchants activated out of millions. Validates B2B-first posture. Consumer ACP transaction layer underperforms today. Data point, not competitor. Reinforces our sequencing: infra first, consumer later.
We do not need to win every segment. We need to be the default rail for agent payments across all of them. D
Appendix E·MPP India adaptation thesis E / F-2 · Reference

MPP is interesting. Not for crypto. For stablecoin-to-UPI bridges.

Our reading: the base protocol is crypto-native, but the adaptation path to stablecoins routed into UPI is both real and narrow. Thesis only for six months.
OriginMachine Payments Protocol. Co-developed by Tempo Labs and Stripe. Launched 18 March 2026. HTTP 402-native with formal authentication scheme submitted to IETF. Payment-method agnostic: stablecoins, cards, Bitcoin Lightning. 50+ services integrated at launch (OpenAI, Anthropic, Google Gemini).
Stripe stackVertical integration deepening: Bridge ($1.1B, Oct 2024) + Privy (75M wallets, June 2025) + Tempo + MPP. Most dangerous global competitor to omni-vault architecture. Pine Labs advantage is narrowly India-specific.
Default railStablecoins (USDC primarily). Designed for programmable, on-chain settlement.
India blockerNative crypto rails not approved. RBI position unchanged through 2026.
Bridge pathTempo or equivalent stablecoin issuer operates off-rail, settles into UPI merchant VPAs. User-side remains INR.
Who benefitsCross-border agent flows into India. Global agent pays, Indian merchant receives INR.
Our actionThesis paper by July. Revisit October with legal read.

Where MPP matters for India · segments

The opening is inbound cross-border, not domestic. A global agent orchestrating on behalf of an overseas user, paying an Indian merchant, with settlement landing in INR without a card network in the path.
Inbound travelGlobal agent books hotel, tour, domestic flight in India. MPP reduces cross-border fees dramatically.
Diaspora remittance-likeNRI agent pays family merchants (school fees, groceries). Agent-native, not person-native.
Cross-border SaaSGlobal SaaS billing Indian users. Agent-managed subscriptions. Currently card-routed.
Export commerceIndian sellers on global marketplaces receive INR without card-network friction. Our B2B positioning fits.
Thesis only. No product build until October. If RBI stance shifts, we reassess. E
Appendix F-1·Agent collections gap F-1 / F-2 · Reference

The one gap nobody has filled.

Agent checkout (consumer → merchant) is live across competitors. Agent collections (merchant agent collects from consumer agent) is not. This is the primitive Pine Labs is building. See architecture in Appendix F-2.
Company Agent checkout
Consumer → Merchant Agent
Agent collections
Merchant Agent → Consumer Agent
Consent vault Verdict
RazorpayConsumer-led
Live — Claude, Zomato, Swiggy, Zepto. Mastercard cohort. Not shipped. No public roadmap. No vault primitive. Focused on checkout. Collections gap open.
CashfreePSP-first
Live — Cashfree Here. ChatGPT + Claude. Mastercard cohort. Not shipped. No vault primitive. Session-based only. No agent-to-agent primitive.
JuspayOrchestration
Partial — orchestration layer, not merchant-facing. Not in scope. No vault primitive. They orchestrate. Collections is structurally out of scope.
StripeGlobal · developer-first
Live globally — Stripe Agent Toolkit. No India UPI. Not shipped for India. Partial — Stripe Connect, not agent-native. No India UPI presence. Collections gap open.
Pine LabsPublic · us
In build — Agent Checkout v1, June 2026. In plan — Agent Collections, June 2026. ACP + UCP compliant. In build — Omni-vault sandbox, July 2026. First mover on collections + vault. The gap we own.
Collections closes subscription, recurring, and B2B settlement use cases for AI-native commerce. Architecture in Appendix F-2. MoltPe reference implementation covers vault + collections primitives. F-1
Appendix F-2·Tri-rail agent payment architecture F-2 / F-2 · Reference

Three rails. One vault. Nobody else has all three.

Rail 1 is table stakes. Rail 2 is where most competitors stop. Rail 3 is the gap — and our moat.
Rail 1 · Table stakes

Agent Checkout

Consumer → Merchant Agent

Consumer agent browses, selects, and authorizes payment to a merchant agent. Standard ACP pattern. Razorpay, Cashfree, and Pine Labs all building here. Not a differentiator — required to compete.

UPI AutoPay Card tokenization BNPL / EMI ACP compliant
In build · June 2026
Rail 2 · Authorization layer

Vault Authorization

Agent → Omni-vault → Debit

Any agent must authenticate through the Pine Labs vault before debiting. Per-agent, per-merchant rules. Passkey consent. The authorization layer that makes Rails 1 and 3 safe and auditable. MoltPe reference covers this surface.

Passkey consent Per-agent rules Velocity caps FIDO2 compliant
In build · July 2026 sandbox
Rail 3 · The gap we fill

Agent Collections

Merchant Agent → Consumer Agent

Merchant agent initiates a collect request. Consumer agent authenticates and authorizes debit from their pre-funded vault balance. Closes subscription, recurring, and B2B settlement for AI-native commerce. No competitor has shipped this primitive. MoltPe reference covers this surface.

ACP collect request UCP consumer auth Pre-funded balance Cross-merchant
In plan · June 2026 · First mover
MoltPe reference implementation (12 months, 75/75 tests passing) covers Rail 2 + Rail 3 primitives. IP transfer to Pine Labs contingent on Chandan + Sanjeev + legal review. Accelerates vault MVP by 3–6 months . F-2