BNY Mellon, now branded as BNY, has deployed tens of thousands of Agentic Assistants across its global operations as of mid‑January 2026, signaling a shift from experimental AI to an autonomous agent‑based operating model.

The initiative emphasizes Explainable AI and governance, using rigorous Model-Risk Reviews, model cards, and feature importance charts to maintain transparency in regulated environments.

The Eliza 2.0 platform acts as the multi‑agent orchestration layer, enabling reasoning and agency beyond chat, and supports a model‑agnostic ecosystem that can switch among OpenAI’s GPT‑4, Google Gemini Enterprise, and specialized Llama‑based models.

Deployment is designed to cut unit costs in core custody trades and improve regulatory compliance through continuous, automated contract review against global regulations.

Looking forward, BNY plans to evolve agents from reactive to proactive roles, including Predictive Trade Analytics and potential Client Co‑pilots, while tackling data privacy and multi‑tenant security for client applications.

This move could set a precedent for other Global Systemically Important Banks, potentially triggering a broader industry shift toward agentic AI in finance amid regulatory and talent-market dynamics.

The scale of adoption is substantial, with tens of thousands of agentic assistants deployed globally, marking a mature transition to an autonomous, agent‑driven operating model.

The AI stack is multi‑vendor and interoperable, with Nvidia providing on‑premises hardware (DGX SuperPOD with H100) and cloud partners Microsoft Azure and Google Cloud Gemini Enterprise delivering integration and advanced research capabilities.

The AI workforce comprises over 20,000 Empowered Builders who design custom agents and more than 130 Digital Employees, each with system credentials and communications access, effectively functioning as autonomous staff members.