Skip to content

dhirendrasrajput/Bharat-Direct

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Bharat Direct

Finishing what UPI started — an aggregator-free, bank-owned payments rail, and a shared Fraud Intelligence Consortium to close the human gap UPI left open.

Status Spec License: CC BY 4.0 Code License: Apache 2.0 Discussion


India built the hardest 80% of a modern payments system — UPI, a bank-owned NPCI, and central CKYC. Two inefficiencies remain. A private payment-aggregator layer still collects rent between merchants and banks, on rails the network no longer needs. And authorized push-payment (APP) fraud — where a person is manipulated into authorising a real payment — is the fastest-growing financial crime, and the one gap a faster rail does nothing to close.

Bharat Direct is an open blueprint to fix both — and an open fraud-intelligence layer being built in the open, right now.

This is a flag, not a pitch for sale. The blueprint is meant to become shared public infrastructure. The author can architect it and is building the fraud layer openly; owning national-scale infrastructure needs NPCI, the banks, RBI, or an anchor player. If that could be you — the blueprint is yours to take.


The 30-second version

  • Pillar 1 — Direct Settlement Consortium. Merchants connect directly to banks; funds settle bank-to-bank over a thin, neutral switch that never holds money; only identity (CKYC) stays central. International cards keep working through one co-owned, at-cost gateway, so merchants never hold card data and stay out of PCI scope.
  • Pillar 2 — Fraud Intelligence Consortium (FRC). A shared, privacy-preserving (federated-learning) risk graph that powers an on-screen Aware Layer — precise warnings delivered at the one moment that matters: the instant of authorisation. Fraud is a network crime; only a shared view sees it whole.

Read the reasoning, the sourced numbers, and the psychology behind the Aware Layer in the white paper.


Start here — pick your door

If you are… Read this Why
A policymaker / regulator (RBI, DFS, NPCI) Concept NoteWhite paper The opportunity, the ask, and the sourced facts in two pages
A bank / anchor player (Reliance, Tata, a consortium) Concept NoteDeck The strategic, category-defining version
A builder / engineer SPEC.mdMANIFESTO.md The open protocol, conformance profiles, and how to integrate
Press / public Op-edCarousel (PDF) The argument in plain language
Just curious MANIFESTO.md The whole idea in one short read

Repository contents

README.md            ← you are here
MANIFESTO.md         ← the idea in one read; the call to the ecosystem
spec/
  SPEC.md            ← the open protocol: pay-by-bank, risk hooks, split settlement, conformance
research/            ← evidence base BEFORE architecture
  RESEARCH-DOSSIER.md← landscape review: DPIP, MuleHunter, TMNL, COSMIC, CIFAS, AFCX, PETs, evidence
fic/                 ← Fraud Risk Consortium Stage-0 reference (PSI, federated) — illustrative
  demo.py            ← two banks catch each other's mules with zero data shared
  PILOT.md           ← how a bank pilots it in a day, with zero real data
docs/
  Bharat-Direct-Whitepaper.docx   ← full discussion paper + sourced data + precedents
  Concept-Note.md / .docx         ← 2-page forwardable note for decision-makers
  Op-Ed.md / .docx                ← ~600-word op-ed
  Bharat-Direct-Deck.pptx         ← pitch deck
  Bharat-Direct-Carousel.pdf      ← square LinkedIn carousel
scripts/             ← the Python generators for the documents above

For builders — the integration promise

The protocol is designed so contributors and consumers integrate without ripping out their core systems: one API, declared conformance profiles (Merchant / Split / Aware / FRC-node), federated by design, easy in and easy out. Start against the mock FRC and the Aware Layer reference, prove the fraud-reduction number on your own rails, then plug into the shared graph. See SPEC.md.

Roadmap

Stage What Status
0 Shared mule-account reporting hub (PSI, federated) Runnable reference in fic/
1 Reputation graph + risk API + Aware Layer Spec drafted; reference WIP
2 Federated models (no raw-data sharing) Designed
3 Golden-hour freeze + shared liability Designed (needs ecosystem)
4 Direct settlement rail + native split Blueprint (needs NPCI / anchor)

Get involved

  • Build / critique: open an issue — cite the section, the actor, and the failure mode. See CONTRIBUTING.md.
  • Partner: banks, PSPs, fraud teams, researchers, and anchor owners are especially welcome — see Contact below.
  • Carry it forward: if you're at iSPIRT, NPCI, a bank, or a player big enough to anchor this, the blueprint is yours to take.

📫 Contact / reach me

If you can see merit and want to convene a conversation — even one I'm not in — please do, and tell me where I'm wrong.


License & status

  • Specification & documents: CC BY 4.0 — share and adapt with attribution.
  • Reference code (scripts and future implementations): Apache-2.0.
  • Status: working draft, released for examination and critique. Figures marked secondary in the white paper should be pinned to primary RBI/NPCI sources before formal circulation. This is a discussion starter, not financial, legal, or investment advice.

About

An aggregator-free, bank-owned payments rail + an open fraud-intelligence layer for India. A flag, not a product.

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages