A recent column in The Ken’s Zero Shot floated a provocation worth sitting with: the super-app dream that Tata and Adani spent years and serious money chasing may finally show up in India — not as another bundled app, but through AI agents. We’re crediting The Ken for that framing without reproducing its paywalled work; what follows is our own India-grounded analysis of why the idea is more interesting than it first sounds, and where it could still fall apart.
The case is counterintuitive precisely because it doesn’t promise a better super app. It argues that agents win by doing the opposite of what super apps tried to do. To understand why that matters, you have to first understand why the super-app playbook never travelled to India the way its champions assumed it would.
The super-app dream India never bought
The blueprint everyone copied was WeChat. Tencent’s messaging app grew into payments, social, mini-programs, and an entire walled economy that Chinese consumers genuinely struggle to live without. For a generation of conglomerate strategists, that became the holy grail: one app, one login, every daily need handled inside a single ecosystem you own and monetise.
The appeal to a large business house is obvious. A super app promises a captive audience, a unified view of the customer, and the ability to cross-sell across verticals you already operate. For groups like Tata, Adani and Reliance — each sitting on retail, travel, payments, energy and infrastructure assets — stitching it all into one front door looked like a natural extension of physical scale into digital life. Gautam Adani, per The Tech Portal’s reporting, reportedly described the ambition as building “the Ferrari of the digital world.”
The Ken’s column takes that same history and asks a sharper question: if the centralised super app didn’t work, does the underlying need it was solving — convenience across many services — finally get met by software agents that act on your behalf? That reframing is the spark for this piece. But before we get to agents, the post-mortems deserve an honest look.
The post-mortems: Tata Neu and Adani One
Adani One is the cleanest cautionary tale. Launched in 2022 as a digital twin of the group’s airport business, it expanded ambitions into payments, ONDC-powered shopping and lifestyle utilities. According to The Tech Portal, citing Bloomberg, Adani invested around $100 million over two to three years. The long-term goal was up to 500 million users by 2030. As of March 2024, it had reached roughly 30 million — meaningful traction in absolute terms, but nowhere near the trajectory needed to justify the spend or build a credible revenue stream.
The outcome, per the same reporting, was a quiet retreat: the digital unit, which had run independently under Adani Digital Labs, was folded back into the airports business, several senior executives departed, and the project reportedly became the subject of an internal mismanagement inquiry. The group is said to be refocusing on infrastructure, energy and logistics — the things it knows.
Tata Neu is the other marquee stumble. Launched in 2022 and boosted by heavy IPL-season promotion, it saw an initial download surge. The momentum didn’t hold. The Tech Portal reports downloads slipped roughly 80% year-on-year in late 2024, with monthly active users falling around 20% in early 2025. Tata Digital, per industry reporting, carried heavy losses through this period. Reliance’s MyJio sits in similar territory — wide reach off the back of telecom, but criticised for uncoordinated features and weak retention.
The pattern across all three is consistent: deep pockets, strong brands, real distribution — and still no daily habit.
Why super-apps failed here
Several failure modes show up again and again, and they’re worth separating because they have different fixes.
- Technical strain at scale. Cramming travel, payments, shopping and loyalty into one shell tends to produce a heavier, slower, glitchier experience. When an app lags, users don’t persevere — they fall back to the single-purpose apps that already work.
- Bolted-together UX. Many of these apps felt like a portfolio of acquisitions stitched behind one login rather than a coherent product. Navigation was confusing, design languages clashed, and the “everything” promise translated into clutter rather than convenience.
- A market already won. India’s app habits are locked in. PhonePe and Google Pay own payments, Swiggy and Zomato own food, WhatsApp owns messaging, and Blinkit, Instamart and Zepto own quick commerce. A new super app isn’t entering empty space — it’s asking users to abandon category leaders they’re perfectly happy with.
- No daily-habit anchor. This is the deepest issue. WeChat earned its place with high-frequency, emotionally sticky hooks — voice messaging, then the lottery-like Red Packets that, by Finshots’ account, added millions of users within days of the 2014 Lunar New Year. Adani One, by contrast, launched on travel: high value, but low frequency. You don’t open an airport app every day. Without a daily reason to return, no amount of cross-sell ambition compounds.
Put bluntly, the Indian super app tried to centralise services that consumers had already happily decentralised — and offered no compelling reason to recentralise them.
Enter AI agents: reversing the logic
This is where The Ken’s provocation gets sharp. The agentic model doesn’t try to fix the super app. It inverts the premise entirely.
As PYMNTS’ Karen Webster argues, smart agents are not the next evolution of super apps — they reverse the logic that made super apps powerful. A super app centralises choice inside a platform: discovery, identity, wallet and the rules of engagement all belong to the operator, and merchants pay for placement to be seen. An AI agent does the opposite. It acts for the user, across many merchants and platforms, organised around the consumer’s intent rather than any single platform’s inventory.
The interface shift is the whole point. In the super-app world, the consumer opens the door and walks into someone else’s curated ecosystem. In the agentic world — to borrow Webster’s framing — the consumer sends software inside on their behalf. You state a goal; the agent interprets it, searches broadly, compares honestly, weighs your stated preferences and constraints, and executes — booking, paying, verifying identity, and managing changes afterward, all within guardrails you set.
That solves the super app’s structural problem without solving its product problem. Super apps failed partly because they needed you to consolidate your habits into their shell. An agent doesn’t ask you to switch apps at all. It works across PhonePe and Google Pay, across Swiggy and Zomato, across whatever airline and hotel fit the brief. The fragmentation that killed the super app becomes the agent’s raw material.
Can agents succeed where apps failed?
That’s the optimistic reading. Here’s the honest balance sheet.
The bull case. Agents sidestep the switching-cost problem because they don’t demand loyalty to a single ecosystem — they orchestrate across many. For the first time, consumers get the same optimisation tools that retailers have used against them for years: an agent that continuously tests price, availability, reliability and service quality across merchants. And the friction agents remove — the searching, comparing, form-filling and deciding — is exactly the friction super apps promised, but failed, to eliminate.
The bear case. Adoption is slower and messier than the narrative suggests. Industry surveys in 2026 point to a familiar gap: a large share of firms are experimenting with agentic AI, but only a minority have scaled it. The realistic path is staged — specialised agents for narrow tasks (travel, reordering groceries, paying bills) arriving before any credible “one master agent.” That’s the same trap super apps fell into, in reverse: trying to do everything at once is how you do nothing well.
And then there’s the fork Webster identifies. Agents face their own defining choice. On one path, they act as genuine consumer fiduciaries, optimising for outcomes you care about and earning money in ways that keep incentives aligned. On the other, they quietly become a new super app — funded by merchant incentives, ad shares and pay-to-play placement, dressed up as neutral advice. If agents go the second way, they reproduce the exact opacity that made platform-curated discovery untrustworthy.
Trust, liability and payments. The hardest questions are operational. When an agent moves money and makes binding commitments on your behalf, who is liable when it books the wrong flight or gets defrauded? Existing consumer-protection and disclosure frameworks weren’t written for autonomous software spending your money. As Webster notes, it isn’t even clear whether agents that transact should be held to standards closer to brokers or advisors. The platforms that win will be the ones designed for auditability — override controls, clear records of the tradeoffs made, and transparency about how recommendations are funded.
The India lens
India has one structural advantage the agentic story rarely accounts for: rails. UPI and the broader Digital Public Infrastructure stack already provide standardised, low-cost, interoperable plumbing for identity and payments. An agent’s hardest execution problem elsewhere — actually moving money across many merchants — is comparatively solved here. UPI is, in effect, ready-made agent infrastructure, and the emergence of agent-aware payment flows could make India a natural proving ground.
Language is the second factor. Agentic interfaces are voice- and text-native, which maps neatly onto a market where the next hundreds of millions of users are more comfortable in Hindi, Tamil, Telugu or Marathi than navigating dense English-first app menus. An agent that takes a spoken instruction in a regional language and executes across merchants could reach Bharat in a way that cluttered super-app dashboards never did.
Regulation is the counterweight. India’s consumer-protection regime, data-protection rules under the DPDP Act, and the RBI’s typically cautious posture on autonomous payments all mean that agents transacting on a user’s behalf will draw scrutiny — over consent, mandates, liability and grievance redressal. That’s not a reason for pessimism; it’s a reason to expect the rollout to be governed, staged and Indian in character rather than a copy-paste of any Western or Chinese template.
What it means for incumbents and founders
The strategic lesson from the super-app graveyard is precise: own the workflow, not the app shell. Adani and Tata spent heavily building front doors nobody wanted to walk through. The value in an agentic world doesn’t accrue to whoever owns the most real estate on a home screen — it accrues to whoever reliably completes a task the user cares about.
For incumbents, that means a shift in posture. Winning is no longer about being the loudest brand or the biggest advertiser inside a platform; it’s about being the best option when an agent evaluates price, availability, reliability and post-purchase experience on your behalf. That requires becoming agent-ready: structured product data, dependable APIs, machine-readable policies, transparent service levels, and post-purchase signals an agent can observe and learn from. A merchant invisible to agents will, in time, be invisible to customers.
For founders, the openings are clearest at the edges:
- The intent layer — vernacular, voice-first agents that translate messy human requests into structured goals for Indian use cases.
- The trust layer — verification, mandate management, audit trails and dispute handling built on UPI and DPI rails.
- Vertical agents — narrow, genuinely useful agents for travel, reordering, bills or healthcare that earn the daily habit super apps never could, before anyone reaches for a master assistant.
The honest conclusion is that AI agents are not guaranteed to succeed where bundled apps failed. They could just as easily collapse into pay-to-play recommendation engines, or stall in the experimentation phase that surveys already flag. But the structural reason for optimism is real: agents don’t fight India’s fragmented, habit-locked app landscape — they exploit it. The super app asked users to change. The agent changes nothing about how you live, and quietly does the work for you. If that promise holds without selling out the user’s interests, the dream Tata and Adani couldn’t buy may arrive precisely because nobody is trying to build a super app at all.
