Topics
AI in Action
This column series looks at the biggest data and analytics challenges facing modern companies and dives deep into successful use cases that can help other organizations accelerate their AI progress.
More in this series
Carolyn Geason-Beissel/MIT SMR | Getty Images
Summary:
Vanguard Group, one of the largest asset management companies in the world, is using artificial intelligence for personalizing client services, creating efficiencies, avoiding costs, and providing better investing advice at low cost. The economic value so far includes shareholder value creation, risk reduction, and operational efficiency. Here, the company’s CIO and chief data analytics officer describes some of the few dozen AI cases Vanguard has underway.
Vanguard Group, an asset management company with a mere $11 trillion or so in assets under management, is a rather successful organization. It’s been in business for 50 years as of this year, but its growth over time has been propelled by an unusual combination of factors for a financial services firm.
Vanguard is owned by its investors. (Tom is one of them, as are more than 50 million other people.) It pioneered the idea of low-cost investing as a means to long-term wealth generation. Tom confesses to being a fanboy (fan-old-man?) for Jack Bogle, the late founder of the company. Tom even holds on to an email he received from Bogle in 2012 after praising Vanguard in an article about Wall Street culture at that time.
Bogle passed away in 2019, but we’re guessing he’d approve of how Vanguard is using artificial intelligence in its business. Vanguard AI is a tool for personalizing client services, creating efficiencies, avoiding costs, and providing better investing advice at low cost. The company’s AI leaders — CIO Nitin Tandon and chief data analytics officer Ryan Swann — keep close track of the value received from data, analytics, and AI, and the return is in the $500 million range thus far. The economic “buckets” for this value include cost avoidance, shareholder value creation, risk reduction, and operational efficiency.
Vanguard is devoted not only to its investors but also to its employees, referred to as “crew,” and AI is being used not to replace crew members but to augment their work. Swann and his colleagues created the Vanguard AI Academy to train crew members on AI, and 50% of them have already completed the program.
Vanguard CEO Salim Ramji — who took over the job in 2024 and is the first incumbent to have come from outside the firm — is also a believer in AI. He is confident that Vanguard is well positioned to innovate with AI, having transitioned its personal investing technology platform almost totally to the cloud in the past few years. Tandon and Swann shared this view when we spoke with them.
AI Use Cases at Vanguard
Ramji said in an interview a few months ago that Vanguard is piloting “a few dozen” AI applications but is holding back on introducing most of them until the kinks have been worked out. Tandon and Swann shared details on some of those cases that Vanguard has already implemented or is piloting:
- A capability for contact center representatives, called Crew Assist, is one of the first Vanguard initiatives using Azure OpenAI. The contact center crew can pose a question to the generative AI tool and get an answer drawn from Vanguard’s internal content. Vanguard receives enormous volumes of calls from its investors, and speeding up resolution time and improving response quality are its primary objectives. The Crew Assist application also speeds new crew member onboarding time.
- Vanguard’s Client-Ready Article Summaries help the more than 150,000 external financial advisers who offer Vanguard’s investment products to provide made-to-order services. The application allows advisers to generate customized summaries of Vanguard’s most-read market perspectives that can be tailored to their customer’s level of financial sophistication, life stage, and preferred tone. The summaries also include necessary disclosures.
- New AI tools for use before, during, and after client phone calls are freeing up time for human advisers. Vanguard has both a robo and hybrid advice service for wealth management, and even the lowest-cost service provides access to human support. The company believes strongly in the value of advice and wants to make client interactions as productive and effective as possible. Advisers can spend up to 50% of their time preparing for calls and summarizing client interactions, which reduces the opportunity for meaningful client engagement. Vanguard launched its first call summarization tool to decrease administrative time and unlock deeper client focus, and more innovations are coming, like automated emails, recaps, and agenda suggestions. Vanguard is also looking to develop real-time, on-call capabilities that could prompt advisers to mention, for example, what Vanguard’s chief economist recently said that’s relevant to the client’s situation, as well as other contextual coaching tips.
- The company’s Digital Advisor provides wealth management services for investors of only $100 or more. This service uses AI to help lower-income clients plan for emergencies by projecting job loss risk and unemployment duration based on historical labor data. In addition, there are GenAI projects in development that would provide these investors with direct AI chats, assistance with financial planning based on Vanguard’s perspectives, and real-time market updates.
- The firm is employing AI-assisted code generation. Like many organizations today, Vanguard’s programmers and system developers are using AI’s code-generation capabilities to increase their productivity. Vanguard analysis found that it provides a moderate level of productivity improvement: around 25% for coding alone, and 10%-15% for the overall system development life cycle. Tandon and his colleagues are still evaluating how to harvest the increased productivity and evaluate the quality of AI-assisted code.
- Vanguard uses AI-based investment approaches. It is incorporating tools such as natural language processing and machine learning-based predictions in some of its active equity quantitative models. Recently, when company analysts conducted an experiment with a large language model to analyze 22,000 company earnings calls, they found that the LLM was effective beyond the traditional quantitative model at predicting dividend activity. For example, companies that the LLM scored as having a “negative outlook” on its calls were almost five times more likely to cut dividends within the following month than companies not rated as negative. That information could be used to eliminate such companies from a portfolio recommendation.
Now that Vanguard is employing a variety of language models across the business, it has adopted a set of systems that evaluate model performance, detect factors like bias and drift, and monitor how crew members are using AI across the firm. Tandon and Swann said that these guardrails are a part of Vanguard’s overall AI governance approach to ensure not only high model performance but also strict adherence to ethical principles and governance policies.
Tandon said that Vanguard envisions advanced AI as “an invisible hand that elevates every facet of our client experience — from how clients engage with us to the results they achieve.” By taking on the heavy lifting of analysis and routine tasks, AI is giving the company’s experts the power to focus more deeply on the financial conversations that clients value.
Topics
AI in Action
This column series looks at the biggest data and analytics challenges facing modern companies and dives deep into successful use cases that can help other organizations accelerate their AI progress.
More in this series
About the Authors
Thomas H. Davenport (@tdav) is the President’s Distinguished Professor of Information Technology and Management and faculty director of the Metropoulos Institute for Technology and Entrepreneurship at Babson College, and a fellow of the MIT Initiative on the Digital Economy. His latest book is The New Science of Customer Relationships: Delivering the One-to-One Promise With AI (Wiley, 2025). Randy Bean (@randybeannvp) has been an adviser to Fortune 1000 organizations on data and AI leadership for over four decades. He is the author of Fail Fast, Learn Faster: Lessons in Data-Driven Leadership in an Age of Disruption, Big Data, and AI (Wiley, 2021).
