India’s Unicorns Are Mining Their Own AI — and the AWS Playbook Is Back

Somewhere in Bengaluru, a real-estate app is now selling artificial intelligence to insurance companies. That sentence captures a quiet but consequential shift underway across India’s startup economy: consumer-tech unicorns that spent a decade hoarding transactional data are starting to turn the AI they built for themselves into products other businesses can buy.

It’s a familiar story with a new cast. Amazon didn’t set out to invent cloud computing; it was solving its own holiday-season scaling problems when the infrastructure it built internally became Amazon Web Services. Slack began as the internal chat tool of a struggling game studio. The pattern internal utility becomes external platform is one of the most reliable value-creation moves in software. India’s unicorns are now testing whether they can run it.
But the reality on the ground is more uneven, and more interesting, than the “everyone is spinning off an AI company” narrative suggests. It’s worth separating what’s actually happening from what’s merely plausible.

The clearest case: NoBroker turns customer service into a product
The most concrete example is NoBroker, India’s first proptech unicorn. To run its brokerage-free marketplace at scale, the company built an in-house system to handle and analyze a punishing volume of customer conversations its platform reportedly processes around 10,000 hours of transcription a day.

Rather than leave that capability buried inside a real-estate company, NoBroker packaged it as ConvoZen.AI, a conversational-intelligence platform for customer-support teams, and launched it as its first major B2B venture. It offers multilingual voice and non-voice automation, real-time analytics, quality control, and compliance tracking, and it has signed outside clients including Cars24, LendingKart, LeapScholar, and Tata AIG.

The strategic logic is textbook. Consumer marketplaces run on thin, cyclical margins. enterprise software sells recurring, high-margin subscriptions. NoBroker’s leadership has been explicit that the AI pivot is tied to a push toward profitability and an eventual IPO. By building its models largely in-house rather than renting them per-call from external providers, the company also claims a meaningful cost advantage the difference between a feature and a business.
This is the AWS analogy actually working: a tool built to solve an internal headache, matured into something other companies will pay for.

The adjacent case: Practo builds an “AI brain,” but keeps it in-house
Practo, the health-tech veteran, is often lumped into the same story — and it is investing heavily in AI but the shape of its bet is different. Rather than carving out a standalone B2B company, Practo is embedding AI directly into its own marketplace as what it calls an “intelligence layer” for healthcare decisions: systems that power patient discovery, decision-making, and care navigation.
Founder Shashank ND frames the opportunity in terms of nearly two decades of accumulated understanding of how healthcare actually works across patients and providers the data foundation, in his telling, for an “AI brain for healthcare.” The company has hired a global chief product and technology officer to drive the effort, crossed $100 million in GMV in the US, and is positioning the AI work as central to a planned public listing.

The distinction matters. Practo is using AI to make its existing platform more valuable, not to sell a spun-off engine to hospitals and insurers as a separate product.

Why the move is so attractive
Where the shift is real, the incentives behind it are powerful and worth naming plainly:
Margin transformation. Consumer-facing platforms in India often operate on razor-thin margins exposed to macroeconomic swings. Enterprise SaaS, by contrast, offers predictable recurring revenue and gross margins that can run well above the consumer business it grew out of.
Valuation re-rating. An AI tool buried inside a logistics or real-estate company gets valued as logistics or real estate. The same capability, carved into an independent deep-tech entity, can command the multiples that investors reserve for pure-play AI a financial-engineering argument as much as a product one.
Elite ML engineers are not always eager to join a transactional marketplace. A dedicated AI company building foundational B2B infrastructure with equity to match is a far easier pitch