Northern Trust's Onboarding Agent Cut Data Prep From Two and a Half Weeks to Three Days

Northern Trust says an AI agent now prepares client onboarding data in three days instead of two and a half weeks — an 83% cut — with a 96.5% F1 first-pass quality score. The announcement with Microsoft covers clients on Northern Trust Asset Servicing’s Front Office Solutions platform, and it publishes a time figure and a quality figure side by side.

The task is narrow and painful. The agent converts a client’s historical transaction data into Northern Trust’s standardized format, which can mean consolidating years of investment data from multiple sources, particularly for alternative assets. Northern Trust cites its own survey finding that 57% of asset owners name consolidating data from disparate sources as their biggest challenge.

Three design choices matter more than the model. First, the agent integrates with Excel and SharePoint, where analysts review, refine and approve its output — human oversight throughout, in tools the team already uses. Second, a governed intelligence layer preserves approved onboarding knowledge over time. Third, Microsoft improved evaluation scores using only anonymized data, which the release calls an eyes-off approach; client data stayed under Northern Trust’s control. COO Pete Cherecwich said the agent reduces the time needed to prepare and validate client data while maintaining expected controls.

The agent was built jointly by Northern Trust and Microsoft Frontier Company Forward Deployed Engineering teams, and that delivery model is spreading. A day later, IBM and Microsoft announced Imagine Enterprise, a platform focused on regulated industries that pairs more than 6,000 prebuilt IBM Consulting agents and assets with forward-deployed engineers from both companies on qualifying engagements.

Both numbers are company-reported, and the release gives no cost. Still, the shape is one a mid-market firm can copy without a hyperscaler’s engineers on site: pick one data-conversion step with a known baseline, measure first-pass quality as well as time, and keep approval with the analysts who own the result. That is how an agent ships inside a regulated business.