In 1992, Tom Carey had just completed his degree in artificial intelligence and felt certain that AI could change the world. Computing power, however, lagged far behind the remarkable ideas stirring in university laboratories. “We had brilliant concepts,” he remembers, “but we simply couldn’t get the hardware to keep up.”
Fast-forward a few decades, and Carey now finds himself at the forefront of a global fintech enterprise that processes trillions of dollars in financial transactions each day. As corporate vice president and president of global technology and operations at Broadridge, he oversees a vast operation in which the company’s India centres play an instrumental role in designing, testing, and deploying cutting-edge AI and distributed ledger technology solutions.
Broadridge’s history traces back to 1962, when its predecessor helped enable the world’s first electronic trade on the stock market. Since then, the company has grown through a blend of product innovation and acquisitions, evolving into a $6-billion industry leader that underpins critical functions in governance, capital markets, and wealth management worldwide. Although it has been publicly listed for the last 15 years, Broadridge operates largely behind the scenes of financial markets, an essential yet often invisible presence. “We serve tier-one and tier-two financial institutions around the globe,” says Carey. “People might not immediately recognise our name, but they certainly feel the impact of our products every time they trade or invest.”
In India, that impact is concentrated in two major hubs—Hyderabad and Bengaluru—where a workforce of more than 5,000 people, about a third of the company’s global headcount, drives innovation. Heading operations in the country is managing director of Broadridge India, Sheenam Ohrie, who previously had stints at Dell, Infosys, SAP and i-flex. “Ours is the only location where you’ll find every Broadridge offering—governance, capital markets, wealth management—all under one roof. It’s the very best place to see how these technologies fit together. That’s what makes India so special for us,” she says.
Carey recalls how Broadridge, then operating under a predecessor firm, set up an India outpost as early as 1999, well ahead of many technology giants. Over time, that outpost grew, culminating in the 2017 establishment of a Bengaluru centre. “We began modestly,” he explains, “but the potential was undeniable. From a technology and innovation standpoint, the country has all the engineering talent one could wish for. We’ve since expanded steadily, and now India stands at the centre of our research, development, and product modernisation efforts.”
Building trust in AI and embracing DLT
In financial services, adopting new technology often runs headlong into concerns about data security and regulatory compliance. Carey points out that his first task at Broadridge was to develop a safe framework for AI. “Our clients are understandably cautious,” he says. “They don’t want their sensitive market data floating outside secure walls, nor do they want to risk violations of regulations. We knew, from day one, that any AI solution we built had to address these concerns head-on.”
Within this framework, Indian engineers have flourished and have developed consequential AI products. At the Bengaluru centre, teams are particularly proud of BondGPT and OpsGPT. BondGPT helps bond analysts slash hours of manual research by drawing on reliable data sets to answer highly specific queries, while OpsGPT streamlines back-office reporting and reconciliation. “If I’m a bond analyst, I can type in a request for all bonds maturing in five years with a yield over three percent,” explains Carey, “and I’ll get a precise, compliant response in seconds.”
In the financial sector, there is simply no room for mistakes. No bond analyst can use information from a publicly available GenAI service like ChatGPT to execute a trade, it would take them too long to verify where the information is being sourced from. “We have to avoid hallucinations— those nonsense answers that AI can sometimes generate. Our system is designed to verify every response it delivers,” Carey says.
To avoid hallucinations, BondGPT draws on a curated mix of publicly available bond market information, including historical yields, maturity data, and credit ratings from reliable sources such as rating agencies and regulatory filings. In addition, with clients’ explicit permission, the platform is trained on anonymised client data to capture real-world trading patterns and market sentiments. By aggregating these data sets—spanning everything from issuance trends to default histories—BondGPT gains an understanding of the bond ecosystem and avoids giving random, but seemingly plausible, answers.
OpsGPT is equally transformative, says Ohrie, especially in markets pushing towards same-day settlement. She points out that in India, regulators have moved to T+1 and even T+0 in some securities, meaning settlement must occur either the next day or sometimes the same day. “Time is no longer a luxury,” she says. “OpsGPT replaces the need for operators to pore over stacks of data or run multiple reports. They can simply ask the system what they need, such as which trades might fail to settle before cut-off. This technology can reduce disruptions and give our clients the agility they need to handle ever-tighter windows.”
While AI attracts much of the public’s attention, Broadridge’s India centres have also pioneered the use of distributed ledger technology, or DLT. The company has launched a digital repo platform designed to automate the short-term lending market, a crucial mechanism for financial institutions. Historically, repo transactions have been laden with paperwork and prone to errors, but Broadridge’s DLT solution—developed largely in India—provides a single, immutable ledger that all participants trust. “Repos are complicated, but if you can eliminate the administrative friction, you’ve taken a big step forward in liquidity management,” Carey says.
Ohrie says the trickiest part of building these innovations, whether the product automates back-office settlements or leverages AI to advise wealth managers, is the pressure to avoid mistakes. “We are processing ten trillion dollars in daily transactions, but we simply can’t afford slipups. Our development teams in India are acutely aware of the enormous responsibility. We do rigorous domain and technical training to ensure that everyone understands both the engineering challenges and the fiduciary obligations that come with them,” she says.
