The Stack as It Stands

Aadhaar (2009–present): Biometric identity verified at population scale — 1.3 billion+ enrolments, KYC cost reduced from $10–20 to $0.27. UPI (2016–present): 81% of retail digital payments, 10B+ monthly transactions. Account Aggregator (2021–present): consent-based financial data sharing — post-AA, formal credit access rose to ~92%. ULI (2023–present): 64 lenders, 136+ data services, 12 loan journeys. The missing layer: governed intelligence.

What Each Rail Does — and What It Cannot

UPI moves money — it does not decide whether money should move. ULI provides plug-and-play data access for credit — but does not assess risk, explain decisions, or detect bias across 136 data inputs. ONDC opens commerce discovery — but does not personalise recommendations, manage supply chain logistics, or detect fraudulent sellers. AA enables consent-based data sharing — but does not interpret what the data means or act on it within governed boundaries. India's DPI provides the pipes. None of these layers decides, recommends, assesses, explains, or acts autonomously.

49%
UPI's share of all global real-time payment transactions
11.74%
Digital economy contribution to India's GDP (FY22-23)
24
Countries India has signed DPI cooperation agreements with
136+
Data services available via ULI for credit assessment

Why Governance Must Be Built In, Not Bolted On

The RBI has confirmed that ULI does not store customer data or borrower consent — responsibility rests entirely with individual lenders. This means governance is not centralised in the infrastructure. It must be built by each institution. FREE-AI's accountability and "Understandable by Design" sutras mandate that every AI-driven decision be traceable to the deploying entity and explainable to regulators. With 136+ data services feeding automated lending decisions, even small biases can amplify across millions of loan decisions without uniform transparency.

The Layer India Should Build Next

Each DPI layer solved a coordination failure that markets alone could not resolve. The next coordination failure is intelligence. India's enterprises — banks, NBFCs, insurers, healthcare providers, utilities — are all deploying AI independently, duplicating effort, creating fragmented governance, and producing inconsistent outcomes. The intelligence layer is where this convergence happens: an enterprise platform that consumes DPI data through governed channels and deploys AI agents meeting the regulatory, ethical, and operational standards India's frameworks demand.

AI agents can only be as capable as the systems they can plug into. In countries with shared, interoperable infrastructure, agents will amplify its value.

— IMF analysis on India's Digital Public Infrastructure, June 2025

References & Sources

  1. IMF, via Biometric Update, "India Is Leading Example of Digital Infrastructure," March 2026.
  2. Government of India, "India's DPI — Cooperation with 24 Countries," February 2026. pib.gov.in
  3. EY India, "AI-DPI Convergence Analysis — IndiaAI Mission," 2025.
  4. World Economic Forum, "Security-by-Design in India's DPI."