Morgan Stanley Wealth Management: AI for Client and Adviser Experience
At COLLIDE 2023, Morgan Stanley Wealth Management's Atul Dalmia described AI systems for personalized client content, adviser next-best-action and lead matching, with reported lifts in contact, engagement and conversion. His lesson: solve the business problem first and augment what already exists.
By The AI Enterprise Magazine Team4 min read
Key takeaways
- Problem first. Dalmia urged teams to solve for a business problem rather than chase a generative AI use case on request.
- Scaled personalization. The engagement engine used 1,500+ client attributes and 40+ content ideas, and reportedly reached six times more clients with 1.5 times higher engagement.
- Humans plus algorithms. Adviser tools combine model recommendations with adviser judgment rather than replacing it.
- Smarter matching. A lead matching algorithm for clients and advisers showed a 60% lift in market testing, he said.
- Engage partners early. Failures are often caused by not involving legal, risk, compliance and business stakeholders, not by technology.
At COLLIDE 2023 in Atlanta, Atul Dalmia, Chief Analytics Officer for Morgan Stanley Wealth Management, walked through three AI use cases the firm had built for clients and financial advisers. The common thread, he said, was starting with the business problem rather than the technology. Dalmia said the firm had about 18 million client relationships and 16,000 advisers at the time, and that generative AI was being layered onto machine learning systems already in production.
How did Morgan Stanley start with the problem, not the AI?
Dalmia opened by asking the room how many had been told by a boss or board to "get me a gen AI use case." Many hands went up. He called generative AI a once-in-20-years technology but warned that financial firms risk "chasing the solution" rather than working out which problems they cannot yet solve well.
His team's mandate, he said, is to create intelligence rather than the "coolest analytics," and to augment advisers rather than replace them. He described a path from machine learning to scaled, AI-powered client engagement, and now to generative AI. He said Morgan Stanley had worked with OpenAI for about two years, building on the firm's own infrastructure. One early rollout let advisers ask a simple query and get a consumable answer from the firm's research instead of searching articles.
How does AI personalize client engagement at scale?
The first use case was an engagement engine, which Dalmia called Genome. Historically, he said, the firm ran marketing campaigns one idea at a time. The challenge was choosing among about 30 ideas at once, explaining to a client why they received particular content, and keeping the experience consistent across phone, email, text, social and mobile.
The engine combines more than 1,500 client attributes with more than 40 content ideas, captures interactions across channels, and uses a deep learning algorithm to refresh personalization daily. Dalmia said the firm learned that the factors driving engagement differ by idea, and that many predictive attributes are not intuitive. That made explainability a large part of the work.
He reported the firm could contact six times more clients, with 1.5 times higher engagement, and that engaged clients brought in more assets. Generative AI, he said, could move this from choosing among static emails to customizing the content itself.
How do advisers use AI recommendations?
The second use case supports advisers, whom Dalmia described as experienced and data savvy. A next-best-action style engine tells them which clients to engage and how, combining algorithmic recommendations with the adviser's own knowledge. "I just met this person yesterday, the person wasn't in a great mood," he said, describing why an adviser might hold a message back.
Dalmia said advisers who used the AI tools engaged more clients and grew faster, and that warming leads through digital channels made phone conversations more effective. He showed conversion likelihood rising from a baseline as clients were warmed digitally, attended events, had a dedicated adviser and held a stock plan relationship. Generative AI, he said, would help aggregate market information so advisers can spend time on conversations.
Can an algorithm match clients with the right adviser?
The third use case is an in-house lead matching algorithm that pairs individuals, such as digital-only clients whose wealth has grown, with the adviser most likely to serve them well. It uses attributes of both sides and learns from which matches historically converted to a full advice relationship.
Dalmia said the firm saw a 60% lift in market testing. He added that call notes and CRM logs had never been truly mined because of the work involved, and that generative AI tools could now extract the drivers of those conversations and feed them back into the model.
What lessons did he share?
Dalmia offered four. Solve for the problem, not for AI, because an outcome is easier to defend than a large investment with nothing to show. Augment existing use cases and tools rather than building from scratch, which shows ROI along the way. Treat data quality and governance as essential, including keeping data inside the firm's systems.
Finally, he said failures are rarely purely technical. They usually come from not engaging legal, risk, compliance, security and business partners early. And expect most experiments to fail: "We've tried 10 things," he said, and only one or two worked.
Where to learn more
The AI Enterprise Conference 2027 takes place Tuesday, August 31, 2027, at Pier Sixty in New York City, and COLLIDE 2027 returns to Sandy Springs Performing Arts Center in Atlanta on Wednesday, October 13, 2027. Details are at /events/ai-enterprise-conference/ and /events/collide/. Registration is coming soon.
Frequently asked questions
How did Morgan Stanley Wealth Management use AI for clients?
Dalmia described an engagement engine that personalizes content across channels using more than 1,500 client attributes and more than 40 content ideas, refreshed daily.
Does AI replace financial advisers at Morgan Stanley?
Dalmia said the firm believes AI should augment advisers, combining algorithmic recommendations with their knowledge of clients.
What results did Morgan Stanley report?
He reported contacting six times more clients, 1.5 times higher engagement and a 60% lift in market testing for adviser-client matching.
What lessons did he share for other companies?
Solve the business problem rather than chase AI, augment existing tools, invest in data governance, engage risk and compliance partners early, and expect most experiments to fail.
This article is based on remarks by Atul Dalmia, Chief Analytics Officer, Morgan Stanley Wealth Management at COLLIDE Data + AI Conference 2023 on October 3, 2023. More coverage: COLLIDE AI Conference.
Filed under: AI Use Cases & Applications, AI Strategy & ROI, People, Culture & Leadership
