AI for Work

BNY partners with OpenAI to create "AI for everyone, everywhere"

Bank of New York Mellon launched Eliza, an internal AI platform powered by OpenAI, letting over 20,000 employees build AI agents for tasks like drafting communications, summarizing documents, and client servicing. Unlike centralized AI teams, BNY's distributed approach lets front-line staff solve problems directly, improving efficiency and client outcomes. The move reflects the industry shift toward AI agents that complete multi-step tasks, and highlights that enterprise AI success depends on broad distribution rather than a single model.

Key Takeaways

  • BNY has launched Eliza, an internal AI platform built on OpenAI models, enabling over 20,000 employees to build their own AI agents that handle multi-step tasks like drafting communications and supporting client workflows.
  • The approach differs from typical bank AI pilots by distributing agent-building to front-line staff rather than a central data science team, closing the gap between what specialists build and what employees actually need.
  • AI agents go beyond chatbots: they can break goals into steps, pull from multiple systems, and complete tasks end to end, which is why enterprises in 2024–2025 are racing to adopt agent-building platforms.
  • Reported benefits include faster internal efficiency and improved client outcomes, such as quicker responses and fewer errors in document-heavy processes, while employees shift toward judgment and oversight roles.
  • Key challenges to watch: governance and regulatory compliance at scale, clearer metrics to measure impact, and whether broad access translates into daily real-world usage — with other financial firms likely to copy the model if it succeeds.

Bank of New York Mellon (BNY) has rolled out an internal AI platform called Eliza, built on OpenAI's technology, that lets more than 20,000 employees construct their own AI agents. The bank says the platform is already improving operational efficiency and client outcomes across the enterprise. Here is what the rollout involves, why BNY took this approach, and what it signals for other large organizations weighing how to scale AI beyond a handful of technical teams.

What Is Actually Happening at BNY

BNY, one of the oldest banks in the United States, is no longer limiting AI work to a central data science team. Through Eliza, an internal platform powered by OpenAI models, more than 20,000 of its employees can now build AI agents — software assistants that can carry out multi-step tasks such as drafting communications, summarizing documents, or supporting client servicing workflows.

The core idea is simple: instead of asking a small group of specialists to anticipate every use case, BNY is letting the people closest to day-to-day problems design the tools that address them. Employees who understand a process bottleneck can translate that knowledge into an agent without waiting in line for engineering resources.

Why an Enterprise-Wide Approach Matters

Many financial institutions have run AI pilots inside innovation labs or single business lines, with mixed results. The common failure mode is a gap between what a centralized team builds and what front-line employees actually need.

BNY's model attacks that gap directly. By opening agent-building to a broad employee base, the bank multiplies the number of experiments running in parallel. A client service specialist, an operations analyst, and a compliance officer will each spot different opportunities for automation, and each can act on them.

The bank reports that this distributed approach is delivering gains in two areas: internal efficiency, meaning tasks get done faster with less manual effort, and client outcomes, meaning customers receive faster or better service. Those are the two measures most boards and shareholders will ultimately judge AI programs by.

What AI Agents Do Differently From Chatbots

A quick distinction is useful here. A standard AI chatbot answers questions in a single exchange. An AI agent can break a goal into steps, use tools or data sources along the way, and complete a task end to end — for example, gathering information from several systems, drafting a summary, and preparing it for human review.

That added capability is why agent-building platforms have become a focus for large enterprises in 2024 and 2025. OpenAI, along with rivals such as Microsoft, Google, and Anthropic, has been pushing tools that let business users assemble agents from existing models rather than writing code from scratch. BNY's Eliza platform fits squarely into this trend, pairing OpenAI models with the bank's own internal controls and workflows.

What This Means for Bank Employees and Clients

For BNY's workforce, Eliza lowers the barrier to participating in AI adoption. Employees do not need a background in machine learning to contribute; they need domain knowledge and a clear problem to solve. That shift can change how roles feel over time, with repetitive work increasingly handled by agents and human attention moving to judgment, relationships, and oversight.

For clients, the practical effects are likely to show up first in turnaround times and consistency — faster responses to routine requests, fewer errors in document-heavy processes, and more time for staff to focus on complex needs.

For the broader industry, BNY's rollout adds to the evidence that enterprise AI success depends less on any single model and more on distribution: how many employees can actually put the technology to work, and how quickly.

What to Watch Next

Three questions will shape how this story develops. First, governance: with 20,000 people building agents, BNY will need strong review processes to keep quality, privacy, and regulatory compliance intact — a challenge every bank deploying AI at scale faces, given the scrutiny of regulators such as the Federal Reserve and the OCC. Second, measurement: the bank cites efficiency and client outcomes, and more specific disclosed metrics over time would help outsiders judge impact. Third, adoption depth: the number of employees with access is one figure; the share who build agents they use daily is the one that matters more.

BNY is a bellwether here because of its scale and its central role in custody and asset servicing, where document-heavy processes are everywhere. If Eliza's bottom-up model holds up, expect other large financial firms to copy the playbook.

Comments (0)

  1. No comments yet. Be the first to share what worked for you.

Leave a comment

Comments are reviewed before they appear. Your email address is not published.