BNY processes AI-assisted client inquiries 80% faster with Eliza, its in-house AI platform

BNY, a 242-year-old bank that oversees approximately $60 trillion on behalf of clients across 100 markets, handles many transactions that lead to client inquiries, and resolving one could involve a human employee and potentially multiple stakeholders. Client onboarding is complex in a highly regulated industry, and its research element involves document processing and decision-making that can lead to weeks of back and forth. Payments with missing or incomplete instructions needed a time-consuming review before they could proceed.

BNY set up an AI Hub in 2023 and built Eliza, an in-house, multi-agentic and model-agnostic AI platform that incorporates Microsoft Azure, Microsoft Foundry, Microsoft 365 Copilot and GitHub Copilot. Eliza brings governance, learning resources, AI tools and agent building capabilities into one platform for employees. On it, BNY created an agentic workflow for client onboarding that reviews new documents, uses previous documentation, pursues new data and helps make a risk decision, and a digital employee that repairs missing or incomplete payment instructions so payments can proceed without a time-consuming review.

More than ten percent of settlement inquiries were resolved or assisted by AI, which BNY says made processing of those inquiries 80% faster. 25% of new onboardings were AI-assisted this year, which BNY says made onboarding 20% faster on average for those clients. The digital employee that repairs missing or incomplete payment instructions handles over 10% of payment repair issues worldwide. The platform now includes over 300 solutions and 100% of employees are trained to use it.

• Build one governed platform where employees find AI tools, learning resources and agent building capabilities, rather than separate point solutions.
• Apply agents to document-heavy, multi-step processes such as client onboarding and payment repair, where back and forth slows the work.
• Train employees on the platform so adoption is led by the people closest to the work.