Morgan Stanley has deployed an internal agentic AI system called FIXR in one of banking's most accuracy-critical and deadline-driven workflows — profit and loss (P&L) reconciliation — and cut the work in half. The counterintuitive part: it got there by making the system less autonomous, not more, according to details shared at a recent VB AI Impact event.
The system goes beyond straightforward generative AI tasks. Humans stay tightly in the loop, and their decisions are iteratively turned into repeatable rules the system can apply on its own. “It's much more like a co-worker than a copilot,” Morgan Stanley Managing Director Todd Johnson said at the event. He added: “We think that's where the opportunity is to really unlock more complex work in the organization.”
Every trading day, Morgan Stanley’s trade desks handle transactions such as cash equities and debt investments. At the end of each day, controllers must reconcile P&L across the firm’s Finance, Risk, Operations, and Trade Capture systems. FIXR automates parts of this high-stakes process while keeping human oversight at every step, reducing what was once a time-consuming and error-prone job.
This approach stands in contrast to many enterprise AI deployments focused on coding assistants or customer service bots. By targeting a high-risk, complex workflow and deliberately constraining the system's autonomy, Morgan Stanley may be charting a pragmatic path for AI adoption in regulated industries where accuracy is paramount. The firm's focus on iteration — turning human decisions into repeatable rules — suggests a template for AI that augments rather than replaces expert judgment.
The broader lesson for enterprise AI is that less autonomy in critical processes can yield more reliable and immediate productivity gains. As financial institutions grapple with regulatory scrutiny and the high cost of errors, Morgan Stanley's FIXR deployment offers a concrete case study in balancing ambition with control.