Instant answers: India's fintechs harness GenAI

Time of India

a day ago

Bengaluru: Generative AI is rewriting the rules of finance, and nowhere is the upheaval sharper than in India's booming fintech scene. Brokerage chatbots that speak four languages, payment assistants that fix errors in seconds, and back-office engines that close two-hundred-million credit accounts before dawn are steadily stripping cost and friction out of money management—and propelling a new breed of startups onto the world stage.
Tired of too many ads? go ad free now
That momentum explains why Tiffany Bloomquist, who runs the startups business for Amazon Web Services (AWS) across Asia-Pacific and Japan, now touches down in India almost every quarter. Meeting founders in Bengaluru during a recent visit, she said GenAI has quickly become the fastest route to improving customer experience and urged entrepreneurs to weave the technology into every layer of their products.
Three local tail-winds, she said, are fuelling the surge.
F
irst, the govet's India Stack gives developers ready-made digital rails for identity, payments and data-sharing. Second, Nasscom counts more than 240 Indian GenAI startups today, up from just 66 early last year. Third, AWS's ten-week GenAI Accelerator—supported by a $230 million fund—offers up to $1 million in credits, matches mentors and organises brisk "speed connect" meetings with big corporate buyers.
Seven Indian firms have already won places on the current cohort.
What, then, are those firms actually doing?
Stock-broking platform Dhan, serving roughly three million traders, was drowning in know-your-customer (KYC) checks. By training a language model on its own policy documents, it now answers a quarter of those queries automatically, halving average wait times and cutting support costs by nearly a third.
Tired of too many ads? go ad free now
Payment specialist Easebuzz faced merchants who repeatedly rang in with integration snags. Its new assistant, ERA, reads each message, compares it with technical manuals and suggests step-by-step fixes in real time. Redundant tickets have dropped by 80% and responses that once took 20 minutes arrive in seconds.
At banking tech company Zeta, the numbers are bigger still. Every night the firm must post millions of transactions that merchants queue during the day.
A cloud-based processing engine—reinforced with AI routines that predict the heaviest bursts—now reconciles 208 million credit accounts in about 40 minutes, a feat that would once have demanded a nine-figure hardware budget.
Debt-marketplace Yubi uses large-language models to refresh credit scores. Feeding the software a lake of public filings and bank statements lets risk models update in hours rather than days, trimming a third off the time borrowers wait for funding offers.
The common recipe is simple: distil company knowledge into an AI model, surface it through a chat-style interface and run it on servers that expand only when the rush arrives. "We used to automate servers," Bloomquist said. "Now we're automating compliance, advice, and trust. Founders who seize that shift will shape the next decade of finance."