Every technology wave has rewritten the Indian opportunity from the bottom up. AI will continue that tradition further, faster, into corners of the economy that every previous wave left untouched. And this time, the interface itself will be Indian. 

I have been a technology investor in India for over a decade I backed companies when “internet penetration” was wishful thinking, when “mobile-first” was a shrug, and when UPI was a government experiment that nobody in a boardroom believed would matter. Each time, sceptics were wrong in the same way. They measured India’s potential against Western baselines and missed the compounding logic of leapfrogging entirely. AI is the next wave. But from my vantage point in Bengaluru, the American frame for this technology misses what is about to happen here. 

The stack India has already built 

Before talking about AI, it is worth revisiting what India has quietly constructed over the past decade in digital public infrastructure. Aadhaar gives over 1.4 billion citizens a verified identity. UPI processed 228 billion transactions in 2025 and is averaging over 20 billion in a single month, surpassing Visa in global real-time payment volumes for the first time. The Account Aggregator framework enables financial data sharing for credit, putting lending tools in the hands of people and businesses that formal banking previously could not reach. DigiLocker holds verified documents for hundreds of millions of citizens. 

No other country has built anything like this at this scale. And critically, this infrastructure was designed to be open. As a platform for others to build on, not a moat for any single company. AI integrated with this stack is not just a consumer product feature. It is rocket fuel for a public digital infrastructure that is already moving. Every layer of the stack has been waiting for an interface that ordinary Indians can actually use. That interface is now arriving. 

121 languages, 1,369 dialects and why AI changes everything 

India is one of the most linguistically diverse nations on earth. The 2011 Census recorded 1,369 rationalised mother tongues, grouped into 121 major languages, of which 22 carry official constitutional status. Two people picked at random from the Indian population have a 91% chance of speaking different native languages. This is a defining fact of India’s social architecture that every prior technology wave either ignored or failed to solve. 

The internet arrived in English. Mobile apps were designed in English, with Hindi as an afterthought and a handful of other languages as an afterthought to that. Even the best vernacular apps of the last decade required the user to be able to read, navigate a menu, tap the right button, or fill in a form. That requirement, innocuous as it sounds, silently excluded an enormous share of the population. 

Voice as the great equaliser 

Voice AI dissolves that assumption entirely. Not voice as a novelty feature bolted onto an existing app, but voice as the primary interface through which a person interacts with a digital service. No menu to navigate. No form to fill. No literacy required. No script to recognise. Just speech, in whichever of India’s languages or dialects a person happens to call their own. 

Voice AI is the first interface in history that does not require the user to be literate to use it 

Think about what this means. A woman in rural Bihar who cannot read has been, for the entirety of the digital era, dependent on a literate intermediary to access any government scheme, any financial product, any health information available online. Voice AI replaces that dependency with direct access, in her language, at any hour, at near-zero marginal cost. The implications extend across healthcare, banking, insurance, agriculture, legal aid, skilling, and every other domain where information is key. 

This is why I believe voice will be the dominant AI interface in India in a way that has no parallel in the West. In the US, voice assistants are a convenience for people who could perfectly well type. In India, voice is the difference between access and exclusion. The stakes are categorically different and so is the market size that opens up when the barrier falls. 

The sectors that brick-and-mortar never cracked 

Take manufacturing. India's ambition to become a global production hub under PLI schemes collides directly with a workforce skilling crisis. Formal ITI training is patchy, outdated, and disconnected from what a factory floor making EV components or smartphones actually requires. An Indic AI small language model deployed on a workstation tablet speaking the worker’s language, pacing itself to the individual solves what classroom training and printed manuals have failed to solve for decades. 

India’s gig economy offers another example of what is possible with targeted AI solutions. According to NITI Aayog, approximately 12 million gig and platform workers are active today, a number projected to reach 24 million by 2030. They are underserved by financial products, undertrained for advancement, and have almost no formal pathway to higher-earning roles. A voice-first AI assistant accessible on any phone can handle job matching, in-context upskilling, and financial guidance calibrated to irregular income. The gig economy is not a problem to be managed. With the right AI layer, it becomes a platform for mass economic mobility. 

India has approximately 13 million small kirana stores, the backbone of consumption in every Tier II city, town, and village. They run on razor-thin margins, informal credit relationships, and paper ledgers. AI assistants can handle inventory, flag reorder points, navigate distributor negotiations, and connect a store owner to working-capital credit through the Account 

Aggregator framework. It can transform today’s struggling kirana into a better running and economically viable business. 

Will AI create or destroy jobs? Two paths 

The American conversation on AI and employment is about which cognitive tasks get automated and what professions contract. That framing only makes sense in an economy where most workers already do structured, digitised, desk-based work. Of India's 610 million workers, approximately 45% are in agriculture and another 25% in informal, manual and geographically dispersed industrial jobs. For the majority of India’s working population, AI is not a threat to their existing role. It is the first technology that has the potential to speak to them directly in their language, on their terms. 

But two futures are genuinely in play. 

The Bull Case: AI expands the effective size of the economy faster than it displaces workers. Sectors too operationally complex and thin-margin to scale, such as construction, kirana retail, rural healthcare, vernacular skilling, become investable at volume. New categories of work emerge. Indic AI trainers and curators, community health navigators with diagnostic AI, gig skilling coaches, and local model supervisors. The demographic dividend gets its enabling technology, 15 years late but not too late. 

The Bear Case: AI accelerates automation in organised services faster than new roles absorb displaced workers. The hardest hit are semi-skilled, semi-urban workers—people who left agriculture, entered organised employment, and now face a second displacement with fewer safety nets. India’s Economic Survey 2023-24 flagged routine cognitive service roles as facing considerable contraction from GenAI. If gains accrue primarily to highly skilled workers, inequality widens rather than compresses. 

In my view, the Bull Case is the more likely 10-year outcome. The Bear Case could still be in play for the next three years for specific cohorts unless government and industry move deliberately on retraining. The difference between the two paths is not the technology. It is the institutional response and the speed at which Indic AI tools reach workers before displacement overtakes opportunity. 

The bet India has to make 

India has never before arrived at a major technology inflection with genuine infrastructure strength. We have the identity layer, the payments layer, the data-sharing framework, and a public system that touches every citizen. What we have lacked, in every prior cycle, is the interface layer—that is the last mile between a digital system and the person who needs it most. For three decades, that last mile was guarded by a requirement so basic it was invisible: you had to be able to read. Voice AI in Indian languages removes that requirement permanently. It is patient, vernacular, requires no instruction manual, and costs almost nothing to scale 

The opportunity is not the transition for the 150 million Indians who are already digitally fluent. They have found their way. It is the next 500 million who represent the largest untapped market for applied AI on the planet. For the first time in a technology cycle, India is not late to the wave. It will not be easy but there is a generational opportunity to use AI to rethink core barriers to education, jobs, health and prosperity for all Indians. Are we willing to invest and bet on ourselves to take on this challenge? 

(The author is Managing Director at Arkam Ventures. Views are personal.)