From copilots to compounding: How India’s digital public infrastructure can drive AI-led growth

India’s coming AI divide could be between companies that use AI to improve an existing interaction and those that redesign how they reach, serve and retain customers they cannot profitably serve today. A co-pilot that answers questions faster may lower costs. Connected to accurate product information, inventory, payment, delivery and support, it can open a new market.

India’s digital public infrastructure can lower the cost of building such services. ONDC enables participating buyer apps to reach sellers across the network, while BHASHINI makes voice and language interfaces easier to build. These are common rails, not a shared customer database. The rails may be public, but product data, fulfilment quality, customer relationships and accountability remain private sources of advantage. Shared rails lower the cost of entry; they do not guarantee differentiation. Every business using them must still earn an advantage through the quality of its data, execution, and customer accountability.

Make the market bigger

Consider a customer who can describe what she needs but struggles to navigate an online catalogue. A voice-enabled service could help her find suitable products, compare options and place an order in her own language. For a business, that could mean a new customer—not simply a cheaper interaction with an existing one.

ONDC’s Saarthi, developed with BHASHINI, and launched in September 2024, offers a starter design for businesses building multilingual shopping apps, including speech recognition and translation. It is a building block, not proof of commercial success. The business still has to connect the conversation to accurate product information, available stock, delivery and after-sales support. A fluent chatbot promising an unavailable product creates disappointment, not growth.

Make protection work

On September 1, IRDAI issued a consultation proposal for a Public Insurance Registry (PIR), a digital public infrastructure intended to create a consistent view of insurance records while allowing relevant information to remain with the institutions that hold it. It is not yet a live registry; its value will depend on the governance, consent mechanisms and operating choices that follow.

Today, policyholders often navigate fragmented records, complex products, and a claims process that begins only after something has gone wrong. The proposed digital rail could give insurers and policyholders a common, consent-led information layer across the policy lifecycle. That could reduce repeated verification, make service more portable and help insurers identify gaps in coverage. The competitive question for insurers will be whether they redesign the full protection journey around it: clearer advice at purchase, more informed underwriting, proactive risk prevention and claims that are faster, more transparent and easier to challenge.

India’s opportunity is to deliver comparable convenience across participating businesses without requiring one company to own every part. Open connections make that possible. Reliable coordination, clear accountability and a service customers trust make it valuable.

This is also a credible global opportunity for Indian technology providers. India is already sharing digital public infrastructure knowledge with other countries, but the opportunity is not to transplant India’s domestic stack wholesale or export Indian citizens’ data. It is to share open standards, implementation experience and AI-enabled service models that can be adapted locally. Indian firms can build multilingual ordering, claims-management, consent and workflow software for retailers, insurers and distributors in other diverse markets. Each market will require its own languages, regulations and institutions; the transferable asset is the operating model.

India’s DPI diplomacy gives those operating models a wider field in which to be tested, adapted and scaled.

What actually compounds

Compounding begins when a completed service makes the next one easier to deliver. In commerce, that may mean a local-language ordering journey, cleaner product information and proven links to sellers and delivery partners. In insurance, it may mean a more complete view of coverage, fewer repeated checks and better-designed interventions before a claim occurs. What accumulates is not unrestricted access to customer data. It is reusable technology, operating knowledge and customer trust. With lawful, consent-led data use and proper testing, each interaction can make the next service more reliable.

This changes the boardroom question. Retailers should ask which customer they cannot profitably serve today. Insurers should ask where policyholders encounter friction, uncertainty or avoidable loss across the protection journey. In both cases, the owner should be accountable for the complete outcome—not for deploying a chatbot or reporting the number of AI interactions.

In both commerce and insurance, the hard part will often sit outside the model. A retailer must correct inaccurate information and honour a delivery commitment. An insurer must explain a decision and honour a valid claim. Shared infrastructure does not remove individual accountability; it makes the quality of that accountability more visible.

India’s AI edge will need capable models and computing power. But businesses will not secure it simply by buying access to either. The opportunity is to combine shared infrastructure with services that reach further, work more reliably and become easier to improve over time. The question is not how much work AI completed today. It is whether the company has built a service, and an operating model, that makes tomorrow’s growth or protection easier to deliver.

(Suman is Partner, Kearney; Thakur is Manager, Kearney. Views are personal.)