BNY Built Its Digital Workforce Backward — And It’s Working

Why It Matters

By prioritizing a governed AI platform and workforce readiness, BNY mitigated the regulatory “trust tax” and achieved scalable, auditable AI automation—setting a blueprint for financial institutions facing strict compliance and ROI pressures.

Key Takeaways

• BNY launched Eliza platform, unifying Anthropic, Google, OpenAI models.
• Over 1,400 staff completed 40‑hour AI bootcamps in 2025.
• 130+ digital employees now handle payment validation and code repair.
• Trust tax forces auditable, least‑privilege AI actions in regulated banks.
• Platform‑first, workforce‑second strategy cut AI rollout time by 200%

Pulse Analysis

Agentic AI is reshaping how banks handle routine and high‑value tasks, but most institutions rush to deploy autonomous agents without the underlying infrastructure. This shortcut often triggers the so‑called “trust tax,” a regulatory burden that demands least‑privilege access, continuous monitoring, and auditable logs for every AI‑driven decision. BNY Mellon sidestepped this pitfall by first constructing Eliza, a model‑agnostic platform that aggregates Anthropic, Google, OpenAI and other large‑language models under a single governed roof. The platform now supports 97% of the bank’s 50,000 employees and powers more than 160 production AI solutions, delivering a 200% year‑over‑year increase in AI‑enabled services.

The second pillar of BNY’s success was an aggressive workforce upskilling program. In 2025, over 1,400 employees completed 40‑hour live bootcamps, and the bank logged more than 170,000 hours of AI learning across both instructor‑led and self‑service tracks. This broad-based competence enabled a bottom‑up adoption model where one‑third of staff have built custom agents to summarize risk reports, generate client insights, and automate their own workflows. By embedding AI fluency across the organization, BNY ensured that the digital employees would be embraced rather than resisted, turning technology into a productivity multiplier.

Only after the platform and talent layers were mature did BNY launch its digital employees—autonomous agents that validate payments, flag transaction anomalies, and even repair code. Each action is fully auditable, satisfying the stringent compliance demands of a G‑SIB. The result is a measurable boost in financial performance: an 18% adjusted compound annual growth rate, 21% pre‑tax income growth, and a 13% dividend uplift for 2025. BNY’s sequenced approach offers a replicable roadmap for banks seeking trusted, ROI‑positive AI, highlighting that the fastest path to scalable agentic automation is platform first, people second, and agents last.