# UPI Study Pay: product case for outbound student payments

*Last updated 21 September 2026 · Uma Ramanathan*

Independent portfolio project. Not affiliated with or endorsed by NPCI. All figures come from the public sources linked in the evidence log.

## 1. The problem

Paying foreign tuition from India costs families an estimated 3–3.5% in hidden exchange-rate markup, takes 2–5 days, needs several forms, and cannot use UPI, the payment tool the same families use every day. The pain points below are ranked by how well the evidence holds up.

| # | Pain point | What the evidence says | Evidence grade |
| --- | --- | --- | --- |
| P1 | Hidden price | Banks apply 3–3.5% markups on the exchange rate; a family sending ₹30 lakh a year may lose ₹60,000–75,000 ([Wise x RedSeer report](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/)). SWIFT fees add ₹1,000–2,500 ([Inforens](https://www.inforens.com/blog/425/sharmistha-das/best-ways-to-pay-international-university-fees-from-india-in-2026)). | C: vendor-commissioned, one source |
| P2 | Slow and untracked | Bank transfers take 2–5 days ([Wise x RedSeer](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/)). Wrong beneficiary details cause rejections or suspense holds ([Karan Gupta](https://www.karangupta.com/blog/rbi-liberalised-remittance-scheme-lrs-2026-rules-for-sending-education-money-abr)). | C: advisory blogs |
| P3 | Paperwork | Form A2 with PAN is mandatory ([RBI LRS FAQ](https://www.rbi.org.in/commonperson/english/scripts/FAQs.aspx?Id=1834)). Form 15CA is filed online, and a CA certificate (15CB) is needed above ₹5 lakh per transaction; advisers suggest starting two weeks before the deadline ([Karan Gupta](https://www.karangupta.com/blog/rbi-liberalised-remittance-scheme-lrs-2026-rules-for-sending-education-money-abr)). | A for A2 and PAN; C for the rest |
| P4 | Tax uncertainty | From 1 April 2026, self-funded education remittances above ₹10 lakh carry 2% TCS (down from 5%); loan-funded education is nil ([Standard Chartered notice](https://www.sc.bank.in/important-information/others/lrs-amendment-in-tcs/), [Matrix Forex](https://matrixforex.in/blogs/matrix-forex/budget-2026-tcs-on-foreign-remittance-what-every-indian-needs-to-know-before-april-1)). The threshold is cumulative across all LRS transfers in the year, so families across banks lose track. | A: bank notice on the Finance Act 2026 |
| P5 | Too many options, no guidance | In a 2023 Flywire survey of 1,800+ students in nine countries, 85% called education payments a significant stressor, Indian students faced 15+ payment options, and 93% wanted instalments ([Flywire](https://www.flywire.com/resources/perspectives-on-higher-education-payments-from-students-in-india)). | C: vendor survey; India sample size not given |
| P6 | UPI cannot be used | Direct outbound UPI transfers from domestic handles to US, UK or EU bank accounts remain prohibited; the only bilateral person-to-person link is UPI-PayNow with Singapore, capped at ₹60,000 a day ([Unbox Future](https://www.unboxfuture.com/2026/09/the-global-upi-reality-check-how-indias.html), [Tech Times](https://www.techtimes.com/articles/327058/20260909/india-declares-upi-its-export-32-million-nris-3-remittance-fees-one-fix.htm)). | B: press, consistent across two sources |
| P7 | Own experience | See section 8: to be filled from your first-hand experience as a recent student. | D: primary, n=1 |

The global picture points the same way. The RBI Governor cited an average cost of 6.2% for a $200 remittance ([Asian Banker](https://www.theasianbanker.com/updates-and-articles/project-nexus-to-transform-global-payments-going-live-in-2026)), and the Financial Stability Board says the G20 cost and speed targets are unlikely to be met by 2027 ([FSB, October 2025](https://www.fsb.org/2025/10/g20-roadmap-for-cross-border-payments-consolidated-progress-report-for-2025/)).

## 2. Market size and why now

The market is smaller than it was, but still large enough for a pilot: RBI-recorded "studies abroad" remittances were $2.31 billion in FY26, and that figure is a floor, because a vendor study puts total student remittances near $10–11 billion. I could not reconcile the two numbers, so every sizing below uses the RBI figure.

| Measure | Value | Source |
| --- | --- | --- |
| Indian students abroad | 1.2 million in 2025, down 5.7% from 1.33 million in 2024 | [MEA data via ICEF Monitor](https://monitor.icef.com/2025/12/the-number-of-indian-students-abroad-fell-in-2025/) |
| Total student spend abroad (tuition and living) | ₹6.3 trillion, about $70 billion, about 2% of GDP | [NITI Aayog via PIE News](https://thepienews.com/inr-6-3tn-indian-student-spend-abroad-sharpens-focus-on-at-home-internationalisation/) |
| LRS education remittances, FY25 | $2.92 billion, down 16.09% | [Business Standard](https://www.business-standard.com/economy/news/outward-remittances-under-lrs-decline-6-85-percent-in-fy25-125052201550_1.html) |
| LRS education remittances, FY26 | $2.31 billion, down 20.9%, out of $28.98 billion total LRS (about 8%, my calculation) | [Business Standard, 22 May 2026](https://www.business-standard.com/economy/news/outward-remittances-slipped-2-to-28-98-billion-in-fy26-rbi-data-126052201720_1.html) |
| Vendor estimate of student remittances | ₹85,000–93,500 crore a year (about $10–11 billion at ₹85 per dollar, my conversion); 95% goes through banks | [Wise x RedSeer](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/) |
| Destination mix | Canada 510,235 study permit holders and US 363,020 students (2024); Australia 139,720 (Jan–Sep 2025); UK 98,015 study visas (to June 2025); Germany 59,420 | [ICEF Monitor](https://monitor.icef.com/2025/12/the-number-of-indian-students-abroad-fell-in-2025/) |

Why this is worth building now, in spite of the decline:

- **Policy tailwind.** On 8 September 2026 the Prime Minister called global expansion UPI's next phase and named costly SWIFT-era remittances as the target ([Tech Times](https://www.techtimes.com/articles/327058/20260909/india-declares-upi-its-export-32-million-nris-3-remittance-fees-one-fix.htm)).
- **Less tax friction.** From April 2026 self-funded education TCS fell from 5% to 2% above ₹10 lakh, so an automatic calculator is now simple to build and to explain.
- **A shifting mix.** Germany's student count has more than doubled since 2020, while Canada and the US shrink ([ICEF Monitor](https://monitor.icef.com/2025/12/the-number-of-indian-students-abroad-fell-in-2025/)). Hypothesis to test: smaller, more frequent payments suit UPI better than one large annual wire.
- **The decline itself is a caution.** April–August education remittances fell from $2.37 billion in 2021 to $1.0 billion in 2025 ([PIE News](https://thepienews.com/study-abroad-remittances-from-india-at-lowest-level-in-eight-years/)), driven by visa limits, rupee weakness and tighter loans.

Scenario, not forecast: a 5% share of the $2.31 billion floor is about $115 million a year, roughly ₹980 crore at ₹85 per dollar. If families save 2–2.5% on it, that is $2.3–2.9 million a year. That is too small to justify the product as revenue for NPCI, so the case has to rest on public benefit, UPI stickiness with young households, and a reusable rail, and section 7 says so plainly.

## 3. What exists today and the gap

No current option gives a family an all-bank, UPI-authenticated payment with the full price shown up front, automatic LRS and tax handling, tracking to the university, and coverage of the countries students actually choose. Each existing route solves part of that.

| Option | What it does well | Where it falls short for students |
| --- | --- | --- |
| Bank SWIFT wire | Universal reach; the default that 95% of flows use ([Wise x RedSeer](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/)) | 3–3.5% markup, 2–5 days, paperwork, little tracking |
| Fintech transfer (Wise) | Shows fees up front and claims the mid-market rate; says 70% of transfers arrive within 12 hours ([Wise x RedSeer](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/)) | Separate app and KYC; the speed and price claims are the vendor's own |
| University payment platform (Flywire with SBI) | Pays in rupees from SBI net banking, with tracking and LRS compliance built in ([Flywire](https://www.flywire.com/news/flywire-partners-with-state-bank-of-india-sbi-to-digitize-education-payments-from-india)) | Only institutions that use Flywire, and only inside one bank's channel |
| Education loan disbursal | Sent directly to the university; TCS nil | Only for borrowers; 7–15 business days of processing plus SWIFT ([Karan Gupta](https://www.karangupta.com/blog/rbi-liberalised-remittance-scheme-lrs-2026-rules-for-sending-education-money-abr)) |
| UPI today | Trusted, used daily, works with every bank; merchant QR payments in about a dozen countries such as the UAE, Singapore and France ([Unbox Future](https://www.unboxfuture.com/2026/09/the-global-upi-reality-check-how-indias.html)) | No outbound transfer to a foreign bank account; none of the five biggest student destinations appears in the merchant-QR country lists I found |
| RBI cross-border aggregator (PA-CB) | Licensed intermediaries with escrow accounts at authorised dealer banks | The RBI circular covers import and export of goods and services with a ₹25 lakh cap per unit, and does not mention education, LRS or individual payers ([RBI notification](https://rbi.org.in/Scripts/NotificationUser.aspx?Id=12561&Mode=0)); a magazine piece names students as a use case ([The Week](https://www.theweek.in/theweek/business/2026/03/21/revolutionising-cross-border-payments-indias-pa-cb-framework-explained.html)), so this needs a regulator's answer |
| Project Nexus | One connection to many fast-payment systems, targeting settlement within 60 seconds | First phase links Singapore, Malaysia, the Philippines and Thailand; India joins in a later phase ([Asian Banker, 2024](https://www.theasianbanker.com/updates-and-articles/project-nexus-to-transform-global-payments-going-live-in-2026)). It does not cover the US, Canada, UK or Australia. |

Two consequences follow. First, waiting for Nexus does not help students going to the US, Canada, the UK or Australia, so the product cannot depend on it. Second, UPI's advantage is not the overseas leg but the front end: every family already has the app, the PIN habit and a bank account behind it. That is the wedge this case builds on.

## 4. The proposed product

UPI Study Pay (working name) lets a parent or student pay tuition to a verified foreign institution from any bank account, authenticated with a UPI PIN, under today's LRS rules, with the full price locked before the PIN and the paperwork done automatically. It is a front end and a scheme, not a new overseas rail, which is why it can start without waiting for one.

**Design principles**

1. **Price before PIN.** One locked quote shows the rate against a published reference rate, the fee, TCS, and the exact amount the university receives.
2. **Compliance done for the user.** PAN, purpose code S0305, the A2 declaration and TCS are filled or calculated, not asked for.
3. **Pay institutions, not individuals.** A verified registry with a name check removes the wrong-beneficiary and fake-link risks and keeps the compliance question simple.
4. **Track to the university.** Five visible states and a receipt the university accepts.
5. **Built for the family.** The student raises the request; the parent, who usually funds it, approves and pays.

**Operating model.** NPCI sets the scheme rules (price display, service levels, disputes) and runs the UPI front end; certified authorised dealer banks and their overseas partners do the foreign exchange, LRS reporting and payout. The alternatives are a single-bank tie-up, which is fast but locks families into one bank as the SBI and Flywire product does, or waiting for direct instant-payment links, which do not cover the main destinations. The scheme model is the recommendation; it is a hypothesis about NPCI's role that a real team would test.

```mermaid
flowchart LR
  A["Parent's UPI app<br/>PIN and approval"] --> B["Study Pay service<br/>quote, rules, tracking"]
  B --> C["Certified AD bank<br/>INR debit, FX, LRS, TCS"]
  C --> D["Overseas partner<br/>local collection"]
  D --> E["University account"]
  B -.status updates.-> A
```

The request travels left to right; status events flow back to the app from the service.

**Requirements.** Priorities: P0 is needed for the pilot, P1 for the second release, P2 later. Targets are proposals to calibrate in the pilot.

| ID | Pri | Requirement | Fixes | Acceptance test |
| --- | --- | --- | --- | --- |
| FR-01 | P0 | Show an all-in quote before authorisation: rate, spread against the reference rate, fee, TCS, amount received | P1 | Amount debited equals the quote in 100% of payments inside the lock window |
| FR-02 | P0 | Verified institution registry with a name and reference check | P2, P6 | 100% of payouts go to registry accounts; mismatches are blocked |
| FR-03 | P0 | Auto-fill PAN, purpose code S0305 and the A2 declaration from KYC | P3 | User types no more than 4 fields (institution, student ID, amount, payer) |
| FR-04 | P0 | Calculate TCS from the purpose, loan flag and year-to-date LRS use | P4 | TCS shown equals the bank's booked TCS in every test case |
| FR-05 | P0 | Five-state tracker with push alerts: debited, converted, sent, received, receipt | P2 | Status updates within 60 seconds of each event |
| FR-06 | P0 | Receipt with reference number, amount and institution that the university can verify | P2 | Receipt accepted in 9 of 10 pilot universities without a follow-up query |
| FR-07 | P1 | Student raises a fee request; parent approves and pays | P5 | Parent completes a request in under 2 minutes |
| FR-08 | P1 | Rate lock for a fixed window, with a clear expiry and re-quote | P1 | Expired quotes never settle at a different rate without new consent |
| FR-09 | P1 | Instalment plan with reminders before each due date | P5 | Reminder sent 7 days and 1 day before |
| FR-10 | P1 | Auto-file Form 15CA; route to a CA partner when 15CB is required | P3 | No manual upload for 15CA cases |
| FR-11 | P2 | Target-rate alerts and loan-financed flow with nil TCS | P1, P4 | Alert fires within 5 minutes of the rate crossing |

**Open design questions.** A single semester's tuition can exceed the ₹5 lakh per-transaction ceiling that applies to some UPI categories ([Outlook Money](https://www.outlookmoney.com/banking/upi-daily-transaction-limits-raised-to-rs-5-lakh-for-select-categories)), so the pilot needs either a special limit for registry institutions or UPI used only for authorisation while the bank settles by another channel. Cross-bank LRS usage has no public API that I found, so the first release relies on the user's declaration plus the bank's own check.

**Phasing.**

| Phase | Window | Scope |
| --- | --- | --- |
| 0 | Months 0–3 | Regulatory alignment, partner selection, prototype tests with 15 families |
| 1 | Months 4–9 | Pilot with 1–2 authorised dealer banks, one destination country, about 10 universities |
| 2 | Months 10–18 | Instalments, rate alerts, loan flow, two or three more destinations |
| 3 | After 18 months | Direct instant-payment links where they exist, living-expense payments, UPI use abroad where accepted |

## 5. Regulatory and feasibility check

The product is feasible inside existing LRS rules only if UPI acts as the channel that authenticates and initiates the payment, while an authorised dealer bank remains the remitter of record; whether the regulator accepts that is the main open question. What is confirmed from primary or bank sources, and what is not, is separated below.

| Topic | Status | Source |
| --- | --- | --- |
| LRS limit | Confirmed: $250,000 per financial year per resident individual, minors included; "studies abroad" is a permitted purpose | [RBI LRS FAQ](https://www.rbi.org.in/commonperson/english/scripts/FAQs.aspx?Id=1834) |
| Channel and documents | Confirmed: remittances go only through authorised dealer banks; PAN and Form A2 are required | [RBI LRS FAQ](https://www.rbi.org.in/commonperson/english/scripts/FAQs.aspx?Id=1834) |
| TCS from 1 April 2026 | Confirmed by a bank notice on the Finance Act 2026: education loan nil; other education and medical 2% above ₹10 lakh; other purposes 20% above ₹10 lakh | [Standard Chartered](https://www.sc.bank.in/important-information/others/lrs-amendment-in-tcs/) |
| Outbound UPI to foreign bank accounts | Reported as prohibited for US, UK and EU accounts; I did not find the RBI or NPCI circular itself | [Unbox Future](https://www.unboxfuture.com/2026/09/the-global-upi-reality-check-how-indias.html) |
| UPI-PayNow person-to-person | Reported: maintenance of relatives and gifts, ₹60,000 a day | [Unbox Future](https://www.unboxfuture.com/2026/09/the-global-upi-reality-check-how-indias.html), [Tech Times](https://www.techtimes.com/articles/327058/20260909/india-declares-upi-its-export-32-million-nris-3-remittance-fees-one-fix.htm) |
| PA-CB route | Not established: the RBI text covers goods and services and does not name education or LRS | [RBI notification](https://rbi.org.in/Scripts/NotificationUser.aspx?Id=12561&Mode=0) |

**Questions to put to RBI and NPCI before any build.**

1. May a UPI-authenticated instruction initiate an LRS education remittance through an authorised dealer bank, with the bank as remitter of record?
2. When a parent pays for a child's tuition, does it qualify under the education purpose code (S0305), or is it treated as maintenance of close relatives (S1301)? The answer changes the TCS from 2% to 20% above ₹10 lakh, so it decides whether the product is worth using. The purpose codes are listed by [Karan Gupta](https://www.karangupta.com/blog/rbi-liberalised-remittance-scheme-lrs-2026-rules-for-sending-education-money-abr); I have not verified the parent case.
3. Can NPCI set a per-transaction limit for verified institutions above the current ceilings?
4. Is certification of authorised dealer providers under an NPCI scheme acceptable, or must each bank be approved separately?
5. Can a consented view of year-to-date LRS usage be shared across banks, so TCS is right the first time?

This is a product case, not legal advice. A pilot would go through the RBI regulatory sandbox or an equivalent approval route before real money moves.

## 6. Pilot, success metrics and service levels

The pilot succeeds if families pay less, know the price first, and the university confirms receipt within a day, without more failures or support load than the bank wire it replaces. It runs about six months with one or two authorised dealer banks, one destination country and roughly ten universities. Every target is a proposal to be calibrated, and where I found no baseline the pilot has to measure one first.

| Metric | Baseline today | Pilot target |
| --- | --- | --- |
| All-in cost (spread plus fee) | 3–3.5% markup plus ₹1,000–2,500 fee ([Wise x RedSeer](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/), [Inforens](https://www.inforens.com/blog/425/sharmistha-das/best-ways-to-pay-international-university-fees-from-india-in-2026)) | 1.5% or less of the amount |
| Time until the university confirms receipt | 2–5 days | 24 hours or less for 90% of payments |
| First-attempt success rate | Not found; measure in the pilot | 97% or more |
| Started-to-PIN completion | Not found; measure | 70% or more |
| Wrong-beneficiary incidents | Reported as a common failure, no rate found | Zero (registry check) |
| Support contacts per 100 payments | Not found; measure | Fewer than 5 |
| TCS errors | Not found; measure | Zero |

On a ₹30 lakh annual payment, a 1.5% all-in cost is ₹45,000. The vendor study puts the hidden loss at ₹60,000–75,000 for the same amount, so the saving depends on how the partners price, which is still open.

**Monitoring and service levels (job description items 5–7).** The pilot team would watch one dashboard in Kibana or a similar tool and hold each partner to written service levels.

| Service | Measure | Proposed level | If breached |
| --- | --- | --- | --- |
| Quote API | Availability, monthly | 99.9% | Incident review within 2 working days |
| Quote API | p95 latency | 1.5 seconds or less | Alert to partner tech lead |
| Payment initiation | Technical decline rate | Under 1.5% | Root-cause report per decline code |
| Status callbacks | Delivered within 60 seconds | 99% | Page the on-call contact |
| Payout to institution | Confirmed within 24 hours | 95% | Escalate to partner operations |
| Reconciliation | Unmatched entries | Under 0.1% | Freeze new volume until cleared |
| Disputes | First response | 4 working hours | Escalate to the scheme owner |

Escalation runs from the operations desk to the partner's technical lead to a senior sponsor. A short daily standup with IT, operations and project delivery reviews yesterday's dashboard, open breaches and blockers, and a weekly note goes to leadership with the same numbers.

Who pays for the service is unresolved: partners would earn their margin inside the 1.5% ceiling, and I have no basis for an NPCI scheme fee. That is a question for the pilot's commercial workstream.

## 7. Risks, counter-arguments and kill criteria

The strongest objections are that the market is shrinking, that fintechs already serve it, and that the regulator may say no; the case survives the first two as a pilot and lives or dies on the third. Each risk below has a response and a result that would stop or redirect the project.

| Risk or objection | Response | Stop or redirect if |
| --- | --- | --- |
| The market is shrinking: LRS education fell 20.9% in FY26 ([Business Standard](https://www.business-standard.com/economy/news/outward-remittances-slipped-2-to-28-98-billion-in-fy26-rbi-data-126052201720_1.html)) | The cost and paperwork problem is per family, not per market; the pilot is small and cheap. The value is public benefit and a reusable rail, not revenue | Interviews show families would not pay a fee at all for tracking and certainty |
| Wise, Flywire and banks already solve it | They solve parts, in separate apps or one bank's channel. The claim to test is that UPI's front end and any-bank access change adoption | Fewer than 30% of interviewed families would switch from a working fintech at the same price |
| The regulator says UPI cannot initiate LRS education payments | Re-scope as a bank-led pilot using UPI for authentication only, or stop | A written no on question 1 in section 5 |
| UPI transaction limits block a semester's tuition | Seek a registry-institution limit, or use UPI for authorisation only | No workable limit and no alternative settlement channel |
| Fraud: fake universities, coerced or mistaken high-value payments | Registry-only payees, name and reference check, first-payment confirmation delay, PIN on every payment | Pilot fraud losses above the agreed threshold |
| Foreign-exchange risk from locked quotes | The partner bears it inside a short lock window; the family never sees a price change after the PIN | No partner accepts that risk at a 1.5% ceiling |
| Destination coverage: partners are needed in each country | Start with one destination; add others once service levels hold | No partner can pay institutions within 24 hours in the first country |
| My evidence leans on vendors, the RBI and RedSeer figures conflict, and my own experience is one case | Section 9 grades every source; section 8 replaces vendor claims with primary research | Primary research contradicts P1 or P2 |

Three results would change my recommendation outright: no partner can deliver at 1.5% or less all-in, the regulator rules out a UPI-initiated LRS payment, or fewer than 30% of families prefer the flow at price parity.

## 8. Validation plan, first-hand experience and delivery plan

The public evidence shows the problem exists; three pieces of primary research over two weeks would show it is yours to solve, and would replace vendor claims with your own data.

**Primary research (two weeks).**

| Method | Sample | What it proves |
| --- | --- | --- |
| Price audit: the same amount and destination quoted the same minute by 3 banks and 3 fintechs, screenshots kept, effective margin over the day's mid-market rate calculated | 6 providers, 2 amounts | Whether the 3–3.5% markup holds without relying on a vendor study |
| Structured interviews about the last payment: price paid, days taken, forms, what went wrong | 15 recent students or parents, mixed destinations and loan or self-funded | Which pain points matter most, and a baseline for section 6 |
| Usability test of the clickable prototype | 8 people | Whether people can state the exact amount the university receives, complete the flow, and prefer it |

> **To do before publishing:** answer the questions below from your own experience, or delete this block. The case marks your experience as evidence grade D (one person) and does not invent it.

**Your first-hand experience (evidence grade D).** Answer these in your own words; they turn the case from research into lived evidence. Leave out anything you would rather not share.

- Which country and how much did you pay, and by which route?
- What did it cost you in total, compared with the mid-market rate on that day, if you checked?
- How many days passed before the university confirmed receipt?
- Which document or step caused the most delay or confusion?
- Did anything fail or need repeating, and how did you find out?
- What would have made you trust the payment more?

**How this case maps to the job description.**

| Job description item | Where this case shows it |
| --- | --- |
| Product design and solutioning | Sections 4 and the clickable prototype |
| Managing technology from a product view | Rails, limits and open design questions in sections 4 and 5 |
| Coordinating with tech and delivery teams | The epics and sprint plan below |
| Regular internal updates | A one-page weekly note: progress, service-level status, risks, decisions needed |
| Agile tools (Azure DevOps, Jira, Slack) | Backlog structure below, ready to load into a board |
| API performance and partner service levels | Section 6 |
| Daily standups with IT, operations, project delivery | Standup format in section 6 |
| Benchmarking and trends | Section 3, including Project Nexus and UPI-PayNow |

**Backlog and first sprints.**

| Epic | Scope | Depends on |
| --- | --- | --- |
| E1 Registry | Institution onboarding, verification, name and reference check | Partner bank data |
| E2 Quote | All-in price, rate reference, lock and expiry | Partner rate feed |
| E3 Compliance | PAN, purpose code, A2, TCS, Form 15CA | Regulator answers, section 5 |
| E4 Payment | UPI authentication, orchestration, limits | NPCI platform team |
| E5 Tracking | Five states, alerts, receipts | Partner status callbacks |
| E6 Monitoring | Dashboard, alerts, service-level reports | E2 and E4 events |
| E7 Family flows | Student request, parent approval, instalments | E1 to E5 |

| Sprint (2 weeks) | Goal | Exit check |
| --- | --- | --- |
| 1 | Prototype tested with 8 users; quote logic specified; registry schema agreed | Test findings logged; open questions sent to the regulator |
| 2 | Quote and rate lock working against a partner sandbox; TCS rules encoded | Quote matches partner value in 100% of test cases |
| 3 | End-to-end payment in sandbox with tracking and receipt | Five states visible within 60 seconds of each event |
| 4 | Monitoring dashboard and service-level alerts live; pilot readiness review | Go or no-go against the kill criteria in section 7 |

## 9. Evidence log

Only pages I opened are listed. Grades: A is a regulator, bank notice or official body; B is established press reporting official data; C is a vendor, adviser or blog whose claims are unaudited. Access date for all: 21 September 2026.

| Source | Used for | Grade |
| --- | --- | --- |
| [RBI, LRS FAQ](https://www.rbi.org.in/commonperson/english/scripts/FAQs.aspx?Id=1834) | $250,000 limit, authorised dealer channel, PAN and Form A2 | A |
| [RBI, PA-CB notification](https://rbi.org.in/Scripts/NotificationUser.aspx?Id=12561&Mode=0) | Scope of cross-border aggregators, ₹25 lakh cap | A |
| [FSB, G20 roadmap progress 2025](https://www.fsb.org/2025/10/g20-roadmap-for-cross-border-payments-consolidated-progress-report-for-2025/) | G20 targets unlikely to be met by 2027 | A |
| [Standard Chartered, TCS notice](https://www.sc.bank.in/important-information/others/lrs-amendment-in-tcs/) | TCS rates from 1 April 2026 | A |
| [Business Standard, FY26 LRS](https://www.business-standard.com/economy/news/outward-remittances-slipped-2-to-28-98-billion-in-fy26-rbi-data-126052201720_1.html) | $28.98 billion total, $2.31 billion education | B |
| [Business Standard, FY25 LRS](https://www.business-standard.com/economy/news/outward-remittances-under-lrs-decline-6-85-percent-in-fy25-125052201550_1.html) | $2.92 billion education in FY25 | B |
| [ICEF Monitor, students abroad](https://monitor.icef.com/2025/12/the-number-of-indian-students-abroad-fell-in-2025/) | 1.2 million students, destination counts | B |
| [PIE News, ₹6.3 trillion](https://thepienews.com/inr-6-3tn-indian-student-spend-abroad-sharpens-focus-on-at-home-internationalisation/) | Total spend, attributed to NITI Aayog | B |
| [PIE News, remittances at eight-year low](https://thepienews.com/study-abroad-remittances-from-india-at-lowest-level-in-eight-years/) | April–August series and drivers | B |
| [Asian Banker, Project Nexus](https://www.theasianbanker.com/updates-and-articles/project-nexus-to-transform-global-payments-going-live-in-2026) | Nexus members and phases; 6.2% average cost quote (2024 article) | B |
| [Tech Times, UPI export](https://www.techtimes.com/articles/327058/20260909/india-declares-upi-its-export-32-million-nris-3-remittance-fees-one-fix.htm) | 8 September 2026 announcement, UPI-PayNow caps | B |
| [Outlook Money, UPI limits](https://www.outlookmoney.com/banking/upi-daily-transaction-limits-raised-to-rs-5-lakh-for-select-categories) | Limits effective 15 September 2025; education not in the new categories | B |
| [The Week, PA-CB explainer](https://www.theweek.in/theweek/business/2026/03/21/revolutionising-cross-border-payments-indias-pa-cb-framework-explained.html) | Claim that students can use PA-CB, which conflicts with the RBI text | B |
| [Wise x RedSeer report](https://newsroom.wise.com/en-CAS/252357-india-s-students-lost-1-700-crores-in-hidden-bank-fees-nearly-three-times-the-cost-of-chandrayaan-3-reveals-redseer-x-wise-report/) | 3–3.5% markup, ₹1,700 crore loss, ₹85,000–93,500 crore flows | C: commissioned by a competitor |
| [Flywire, India student survey](https://www.flywire.com/resources/perspectives-on-higher-education-payments-from-students-in-india) | Stress, option overload, instalment demand (2023) | C: vendor survey |
| [Flywire and SBI announcement](https://www.flywire.com/news/flywire-partners-with-state-bank-of-india-sbi-to-digitize-education-payments-from-india) | What the existing bank product does | C |
| [Unbox Future](https://www.unboxfuture.com/2026/09/the-global-upi-reality-check-how-indias.html) | Outbound UPI limits, merchant-QR countries | C |
| [Inforens](https://www.inforens.com/blog/425/sharmistha-das/best-ways-to-pay-international-university-fees-from-india-in-2026) | SWIFT fee range | C |
| [Karan Gupta](https://www.karangupta.com/blog/rbi-liberalised-remittance-scheme-lrs-2026-rules-for-sending-education-money-abr) | Process steps, purpose codes; its TCS figures are out of date and not used | C |
| [Matrix Forex](https://matrixforex.in/blogs/matrix-forex/budget-2026-tcs-on-foreign-remittance-what-every-indian-needs-to-know-before-april-1) | Cross-check of the April 2026 TCS change | C |

**Not verified.** I did not find the RBI or NPCI circular that bars outbound UPI transfers, NPCI's own list of merchant-QR countries, or the UPI-PayNow terms from NPCI or MAS. Sources disagree on when Project Nexus goes live and on whether UPI is in 11 or 12 countries. The RBI education figure ($2.31 billion) and the vendor flow estimate (about $10–11 billion) are not reconciled.