"You don't need to have deep AI knowledge to build AI-enhanced tools now. It's all about identifying the right use case."
When you think of financial services, the first thing that likely comes to mind isn't innovation. Instead, you might envision the endless paperwork, regulatory red tape, and Excel sheets that seem to stretch on indefinitely. However, things are changing quickly and for the better.
With generative AI agents now capable of acting like hyper-efficient virtual coworkers, even the most bureaucratic corners of finance are getting a serious upgrade.
However, Broadridge's foray into using AI in this isn't new. It happened long before everyone started talking about ChatGPT.
For over a decade, the New York-based technology and infrastructure provider has been experimenting with building its own predictive and analytical models primarily for efficiency, and they have been in place for years. Over time, the AI use cases have matured and scaled.
From parsing unstructured regulatory filings to forecasting shareholder voting outcomes, Broadridge is helping financial institutions integrate AI not just to automate, but to anticipate. Inigo Fernando, the CTO for Broadridge India's governance and communications business unit, told AIM about the journey.
Broadridge powers investing, governance, and communications for its clients, with its operations platforms processing and generating over 7 billion communications per year and underpinning the daily trading of more than $10 trillion of securities globally.
Their governance and communications unit is focused on what Fernando calls "regtech" -- regulatory technology. This is a space where AI is becoming increasingly critical.
The core responsibility of Broadridge is to help publicly traded companies communicate with their investors, something that's not just good practice but a regulatory mandate. "This is part of the 'G' in ESG -- governance," he said.
When it comes to using AI, Fernando believes most of AI is becoming a commodity. "You don't need to have deep AI knowledge to build AI-enhanced tools now. It's all about identifying the right use case."
One of the most impactful applications of AI at Broadridge has been in the space of investor insights.
"Every year, companies hold their AGMs, where investors vote on everything from electing directors to approving mergers. We've developed AI models that help these companies predict voting outcomes in advance -- whether quorum will be met, whether a proposal will pass, and so on," Fernando shared.
Behind the scenes, several gradient-boosting algorithms crunch decades of historical voting patterns, investor behaviour, and industry trends to provide predictive intelligence.
"The models are built using massive volumes of structured data. The real secret sauce isn't just the model -- it's the depth of historical data we have from being in the industry for over 40 years," he added.
"We go through a rigorous evaluation of various models -- security, accuracy, suitability to the data structure. Older AI models worked well for us in this case. But we're always keeping an eye on emerging tools or even more advanced neural networks."
Another powerful example of Broadridge's AI deployment lies in regulatory document simplification. He explained that the SEC introduced a regulation around Tailored Shareholder Reports, allowing mutual funds and ETFs to replace their massive 80-100-page reports with easy-to-read two-pagers.
"For us, that meant taking thousands of unstructured filings, extracting meaningful financial data, and distilling it into a concise, investor-friendly format. We used several LLMs for this. It's a perfect fit when AI reads like a financial analyst and writes like a human."
Predicting the stock market trends is one thing that Fernando believes is possible as well. "I don't think humans would rule that possibility out. I wouldn't rule that possibility out either," he commented.
For example, last year, LTX, the AI-driven corporate bond trading platform backed by Broadridge Financial Solutions, unveiled a GenAI-powered List Trading feature. Expanding its leadership in the fixed income space, the platform now enables users to create and execute complex, multi-asset, multi-directional trade lists using GPT technology through its RFQ+ system.
But not all AI at Broadridge is flashy. Some of it is quietly transforming back-office operations.
Many of the operator tasks are repetitive, time-consuming, and prone to error. Broadridge started introducing AI modules to assist decision-making in these processes. "But we're careful. We always keep a human in the loop to ensure accuracy and compliance."
The company has also experimented with recommender systems for bond trading, helping investors find potential counterparties by analysing historical transaction data using neural networks.
Broadridge plans to double down on LLMs and generative AI in the near future. "We're already using LLMs for things like document summarisation and understanding unstructured data. For structured data predictions, traditional models still work better. But the lines are blurring," he said.
On the other end of Broadridge, at the capital markets, wealth, and asset management business unit, the firm also uses AI tools like OpsGPT, BondGPT, Distribution GPT, and BroadGPT, enabling advanced automation, real-time insights, and enhanced operational intelligence.
Kishore Seshagiri, executive director and chief digital officer, Broadridge India, told AIM that as AI adoption accelerates, the team remains committed to responsible AI integration, prioritising transparency, security, and ethical governance.
For example, BondGPT is trusted by over 700 traders and streamlines bond trading and portfolio management by using OpenAI GPT-4 to answer bond-related queries, enhance dealer-client engagement, and optimise decision-making.
Another example is BroadGPT, which is a productivity assistant, helping over 15,000 associates enhance their work quality by streamlining content creation, translating surveys and reports, assisting with pair programming, and serving as a valuable learning tool.
