Morgan Stanley Wealth Management made AI faster to build without loosening the controls

Across roughly 2,500 active business flows, Morgan Stanley Wealth Management built a governed foundation that lets developers, business users, contact center teams, and executives work with AI and trusted data, while reducing selected flow-development effort from 20–40 hours to just 3–5.

Everyone wanted AI. Scaling it safely was the harder problem.

Across Morgan Stanley Wealth Management, teams were already seeing where AI could help. One of the world's premier financial services firms, Morgan Stanley has long treated rigorous controls not as a constraint on innovation but as the foundation that makes it possible at scale. Developers wanted to build faster. Contact center teams wanted to make sense of client interactions. Business users wanted to ask questions of trusted data without waiting for technical support. Executives wanted a faster path to insight.

The demand wasn't the problem. Making AI broadly available inside a highly regulated financial institution was.

With roughly 2,500 active business flows, Morgan Stanley couldn't treat every new AI use case as a separate experiment. It needed a repeatable approach to security, data isolation, access, cost, and oversight.

Build the guardrails once. Let more teams use them.

Morgan Stanley Wealth Management used Dataiku with Snowflake Cortex to create an enterprise-approved pathway for AI. The same foundation supports different experiences for different users.

Developers can accelerate SQL and code development. Contact center teams can turn call notes, emails, and client interactions into summaries and structured signals. Business users and executives can ask questions of trusted data in natural language.

One AI foundation. Different experiences for the people using it.

Reducing 20–40 hours to 3–5 hours

For selected business flows, AI-assisted development has reduced effort from approximately 20–40 hours to 3–5 hours. Across Morgan Stanley Wealth Management's environment, those productivity improvements contribute to an estimated $3.75 million in annual savings and cost avoidance. But faster development is only one part of the shift.

Business users can increasingly move from a question to analysis without waiting for someone else to build it. Technical teams can spend more time curating trusted data, business logic, and architecture rather than handling every request themselves.

AI isn't replacing expertise. It's removing some of the work standing between expertise and the answer.

Built-in governance accelerates time-to-scale

Broader AI access only works if the controls expand with it. Morgan Stanley Wealth Management built guardrails around flow-level access, data isolation, LLM usage and cost, platform resources, and architecture review. Capabilities must demonstrate that required safeguards are in place before rollout.

That means governance doesn't have to be reinvented every time another team wants to use AI. Instead of governance being the final gate, it becomes the reason AI can scale.

Dataiku was already a mature analytics platform inside Wealth Management, so generative AI became an extension of an existing foundation rather than another disconnected environment.

Now Morgan Stanley can continue expanding natural-language access and AI-assisted development while maintaining the controls a regulated business requires.

The bigger shift isn't simply that more people can use AI. It's that the organization has a repeatable way to introduce it.