BNY and OpenAI Partnership: Scaling Enterprise AI via the Eliza Platform

BNY has partnered with OpenAI to implement a platform-based approach to generative AI, deploying a centralized system called Eliza that allows 20,000 employees to build and deploy AI agents. This initiative has resulted in over 125 live use cases across all major business lines, shifting the bank's operational model from isolated experiments to enterprise-wide integration.

The Eliza Platform: Integrating Governance and Development

BNY utilizes Eliza as a "system of work" that embeds governance directly into the tooling to ensure safety and accountability for a systemically important financial institution managing over $57.8 trillion in assets. Rather than treating governance as a barrier, BNY uses it as an enabler to accelerate deployment.

Governance Structure

To maintain trust and security, BNY employs three primary cross-disciplinary oversight bodies:

  • Data Use Review Board: A cross-functional group focusing on intellectual property, cybersecurity, engineering, data, privacy, and third-party relationships.
  • Artificial Intelligence Release Board: A team that reviews initiatives specifically before they are deployed into production.
  • Enterprise AI Council: A senior body providing firm-wide policy alignment and oversight.

Governance is integrated at the system level within Eliza, meaning that model selection, prompting, agent development, and sharing are all conducted within a governed environment with standardized permissions and security.

Employee Enablement and Cultural Shift

BNY has prioritized workforce readiness, with 99% of its employees trained on generative AI. The firm uses a combination of mandatory training and community-driven learning to empower non-technical staff to act as developers.

Training Initiatives

  • Make AI a Habit Month: A series of daily seven-minute trainings focused on prompting and agent building, which led to a 46% increase in the number of agents created.
  • Cross-Functional Hackathons: Events where employees from Legal, Sales, and Engineering collaborate to build tools, allowing non-IT staff to solve their own business problems.

AI Use Cases and "Digital Employees"

BNY has moved from simple knowledge extraction to the creation of "digital employees"—advanced AI agents with their own identities, access controls, and dedicated workflows.

High-Impact Use Cases

  • Contract Review Assistant: Reduces legal review time for over 3,000 annual vendor agreements by 75%, cutting the process from four hours to one.
  • Risk Insights Agent: Surfaces emerging risk signals across portfolios through deep research to allow analysts to act proactively.
  • Lead Recommendation Engine: Generates client-relevant insights and opportunities for sales teams.
  • People Business Partner Agent: Automates responses to benefits and policy queries to improve consistency.
  • Metrics Agent: Provides permission-aware summaries of learning platform performance.

Advanced Capabilities

Certain teams use ChatGPT Enterprise and its "deep research" capabilities for multi-step reasoning across internal and external data. This is applied to risk modeling, scenario planning, and novel legal evaluations.

Enterprise AI Implementation Blueprint

BNY's strategy for scaling AI in a highly regulated environment provides several key lessons for AI leaders:

  • Extend Existing Frameworks: BNY extended its mature legal and compliance processes to AI rather than building new governance from scratch.
  • Embed Governance in Tooling: By making tagging, telemetry, and approval flows part of the Eliza interface, governance is enforced without adding manual burdens to users.
  • Shared Responsibility: Using cross-functional councils ensures that domain-specific risks are evaluated in real-time.
  • Prioritize Literacy: Ensuring nearly the entire workforce is trained on responsible AI use is a prerequisite for identifying both risks and opportunities.

Sources