Verified
BNY
Piloting- Custody bank
- 46,600 employees
- ~$4B tech budget
- Eliza, built in-house
- 400+ AI solutions in production
- ~140 digital employees
BNY deploys AI through Eliza, an enterprise platform it launched in 2024 on Microsoft Azure that routes work across OpenAI, Google, Anthropic and open-source models, backed by its own NVIDIA DGX SuperPOD. Every BNY employee has Eliza access, about half used AI daily in Apr 2026, and CEO Robin Vince said in Oct 2026 that more than 400 AI solutions run in production, alongside digital employees that repair payments and fix code vulnerabilities. BNY says AI earns a return but does not publish AI spending or savings.
bny.comTheir approach to AI
- The platform
- Eliza, the enterprise AI platform BNY launched in 2024 and moved to Eliza 2.0 in 2025, puts chat, deep research, agent building and AI governance in one place and runs on Microsoft Azure and Microsoft Foundry (Sep 2026).
- The models
- Eliza is model-agnostic: it integrates OpenAI, Google and Anthropic models (2025 annual report), and Robin Vince said in Oct 2026 that BNY also uses open-source models and routes each task to the model that fits; GitHub Copilot and Microsoft 365 Copilot sit inside Eliza.
- The reach
- Every BNY employee had Eliza access and about 50% used AI daily in Apr 2026; CIO Leigh-Ann Russell put daily or super users at almost two-thirds of staff in Sep 2026.
- The catch
- BNY does not break out AI spending or returns: Robin Vince said in Oct 2026 that AI is generating a return and that its cost folds into a $13 billion expense base that includes $4 billion of technology, and BNY publishes operating gains, not dollar savings.
Snapshot
- Employees
- 46,600 full-time
- Sep 30, 2026; 48,100 at Dec 31, 2025, about 60% outside the U.S.
- Technology budget
- ~$4B a year
- Robin Vince, Oct 2026; Dermot McDonogh called it an engineering budget of about $4B in Jul 2026
- AI/ML headcount
- Not disclosed
- AI Hub created in 2023 to bring data science, AI and machine learning teams together
- Who owns AI
- Sarthak Pattanaik, Chief Data and AI Officer
- Named by Aug 2025; CIO Leigh-Ann Russell runs engineering and reports to Robin Vince
- Platform
- Eliza, built in-house
- Launched 2024; Eliza 2.0 in 2025; built on Microsoft Azure and Microsoft Foundry
- Models
- OpenAI, Google, Anthropic, open-source
- Model-agnostic, with routing by task (Oct 2026)
- Reach
- 100% access; ~50% daily users
- Apr 2026; almost two-thirds daily or super users per the CIO, Sep 2026
- Use cases
- 400+ AI solutions in production
- Oct 2026, plus about 140 digital employees (Apr 2026)
Key findings
Robin Vince said on Oct 15, 2026 that BNY has more than 400 AI solutions in production, up from 160 at the end of 2025 and about 220 in Apr 2026, when the bank said every employee had access to Eliza and about half used AI daily.
BNY ran about 140 digital employees in Apr 2026, multi-agent AI systems with their own credentials and a human supervisor, and a digital employee handled more than 10% of BNY's payment repair issues in Sep 2026; American Banker put the count above 150 in Oct 2026.
BNY does not break out what AI costs or returns: Robin Vince said on Oct 15, 2026 that BNY spends $4 billion a year on technology within a $13 billion expense base and that AI is generating a return, while Dermot McDonogh said in Jan 2026 that BNY spends more on cyber resiliency than on AI and in Oct 2026 that BNY now tracks its AI token spending with its own telemetry.
BNY keeps Eliza model-agnostic: the platform runs on Microsoft Azure and Microsoft Foundry, integrates OpenAI, Google and Anthropic models, added Google's Gemini Enterprise in Dec 2025, and uses open-source models and routing to control token costs (Oct 2026), alongside BNY's own NVIDIA DGX SuperPOD.
BNY had 46,600 full-time employees on Sep 30, 2026, down from 49,200 a year earlier, and Dermot McDonogh said in Jan 2026 that the earlier decline was not yet due to AI; Robin Vince joined the boards of the OpenAI Foundation and OpenAI Group PBC in Jul 2026.
Where BNY sits
- Stage
- Piloting
BNY has reported no business return it credits to AI, for any use case or for the firm as a whole: Robin Vince said on Oct 15, 2026 that AI is generating a return, without giving a figure, and that BNY does not break AI out of its expense base; the AI gains BNY publishes are operating measures, such as settlement inquiries investigated more than 80% faster; and BNY does not say how much of its roughly $550 million of 2025 efficiency savings came from AI. BNY has officially reported a company-wide rollout, which places it in Piloting: its Apr 2026 investor presentation said every employee has access to Eliza and about 50% use AI daily, and Chief Data and AI Officer Sarthak Pattanaik owns the program. BNY falls short of Scaling because it reports no AI-credited return in any function and no firm-level AI figure, and it presents its higher revenue per employee as the result of reimagining the company, not as a return from AI.
How the stages are defined - Ownership
- Chief Data and AI Officer Sarthak Pattanaik; CIO Leigh-Ann Russell reports to the CEO
- Platform
- Hybrid: Eliza built in-house on Azure; Copilot tools bought
- Models
- OpenAI, Google, Anthropic and open-source, routed by task
- Hosting
- Microsoft Azure and Microsoft Foundry under Eliza, Google Cloud for Gemini Enterprise, and BNY's own on-premises NVIDIA DGX SuperPOD
- Reach
- All employees have access; about 50% daily users (Apr 2026)
How they deploy
One platform for every employee
BNY built Eliza as a single enterprise AI platform partly to avoid redundant tool investments across teams (Oct 2025), and Robin Vince's 2025 annual report letter calls the central platform approach, drawing on external platforms and open-source capabilities, the right one for BNY at this stage. BNY's stated vision is AI for everyone, everywhere and everything, and the bank made Eliza available to 100% of its employees by Apr 2026.
Model-agnostic, with routing to control cost
Robin Vince said on Oct 15, 2026 that BNY pits models against each other, from frontier labs to open-source and older models, and routes each task to the right one inside Eliza, which he described as the layer that every BNY agent and employee works through. Vince said on the same call that BNY measures token costs precisely and that its partnerships with several AI labs set up better commercial terms, and in Apr 2026 that a firm living inside one lab's ecosystem risks lock-in and token price wars. Dermot McDonogh said in Oct 2026 that BNY built telemetry to track how it spends AI tokens and does not use expensive models for cheap queries.
Capacity first, then products and new services
Robin Vince describes three ways AI pays off at BNY: creating capacity inside the company, improving products for clients, and doing new things for clients that BNY would not have done without AI. Vince said in Oct 2026 that BNY's AI work started with capacity and is shifting toward product improvements, and that BNY sees a chance to extend Eliza's workflow capabilities to clients over time.
From breadth to depth in 2026
BNY's 2025 annual report sets adoption and integration as the priorities for 2026: moving from access and early proof points to AI embedded in end-to-end processes, product development and client delivery. Robin Vince said in Apr 2026 that BNY is moving from AI point solutions to reducing manual touch points, shortening cycle times and strengthening control outcomes across whole workflows.
Digital employees
BNY introduced digital employees in 2025: multi-agent AI systems that work alongside staff, each with a distinct persona, credentials and a human supervisor. BNY's digital employees validate payment details, repair incomplete payment instructions and remediate code vulnerabilities, and BNY counted 134 at the end of 2025 and about 140 in Apr 2026; American Banker reported more than 150 in Oct 2026.
- Deployment model
- BNY runs AI as one shared platform with use cases built across the company. The AI Hub, created in 2023, brought BNY's data science, AI and machine learning teams together, and Chief Data and AI Officer Sarthak Pattanaik and CIO Leigh-Ann Russell both speak for Eliza's design and its Microsoft foundation (Sep 2026). Any BNY employee can build on Eliza, most AI builders come from outside engineering (Oct 2025), and new AI ideas go to a review committee that meets daily (Sep 2026). BNY moved more than 70% of its employees into a platforms operating model by the end of 2025, which Robin Vince said in Jul 2026 gives BNY an advantage in embedding AI.
- Model providers
- BNY's 2025 annual report says Eliza integrates models from OpenAI, Google and Anthropic, and Robin Vince said on Oct 15, 2026 that BNY also uses open-source models. BNY entered a multi-year collaboration with OpenAI in 2025 and in Dec 2025 integrated Google's Gemini Enterprise into Eliza for deep research, using Gemini 3 and Veo 3. Microsoft Foundry supplies models and model lifecycle management, Microsoft 365 Copilot is integrated into Eliza, and software engineers use GitHub Copilot through Eliza (Sep 2026). Robin Vince said in Apr 2026 that BNY is running Anthropic's Mythos model in-house for cyber defense, along with an early-access OpenAI security capability.
- Data and infrastructure
- BNY built Eliza on Microsoft Azure, which Sarthak Pattanaik chose for scalability, resilience and the data BNY already holds with Microsoft, and on Microsoft Foundry for model usage observability, cost thresholds and sovereignty (Sep 2026). BNY also became the first global bank to deploy an NVIDIA DGX SuperPOD in 2024, on-premises GPU infrastructure that supplements its cloud capacity, and Robin Vince said in Jan 2026 that BNY has its own NVIDIA hardware alongside its Google and OpenAI collaborations.
Platforms and internal tools
Eliza2024; Eliza 2.0 in 2025
Eliza is BNY's proprietary enterprise AI platform, named after Eliza Hamilton, wife of founder Alexander Hamilton. Eliza offers secure chat, search, deep research, coding and agent building, a marketplace of AI solutions and approved datasets, and also serves as BNY's AI governance dashboard and learning platform. BNY describes Eliza as model-agnostic and multi-agent, and Eliza 2.0 added a more agentic experience in 2025.
- Users
- 100% of employees with access; about 50% daily AI users (Apr 2026)
- Models
- OpenAI, Google, Anthropic and open-source models
Digital employees2025
Digital employees are BNY's multi-agent AI workers built on Eliza, each with a distinct persona, credentials and a human supervisor, deployed in payments and engineering to validate payment details, repair payment instructions and remediate code vulnerabilities. BNY's 10-K warns that digital employees and digital engineers add operational and control risks, including errors in automated decisions and gaps in oversight.
- Users
- 134 at the end of 2025; about 140 in Apr 2026
Microsoft 365 Copilot and GitHub Copilot in Eliza
BNY was an early Microsoft 365 Copilot adopter in financial services and has integrated it into Eliza, and BNY software engineers reach GitHub Copilot through Eliza. BNY plans to add Microsoft's Work IQ and Copilot Cowork (Sep 2026).
- Users
- Almost 60% of BNY onboarded to Microsoft 365 Copilot (Sep 2026)
- Models
- Microsoft 365 Copilot, GitHub Copilot
Gemini Enterprise in ElizaDec 2025
BNY integrated Google Cloud's Gemini Enterprise into Eliza to extend agentic deep research for market analysis and give BNY's workforce multimodal tools; BNY employees build agents that process and analyze financial reports, data and historical trends.
- Models
- Gemini 3, Veo 3
NVIDIA DGX SuperPOD2024
BNY's on-premises GPU infrastructure, which made BNY the first global bank to deploy an NVIDIA DGX SuperPOD, supplements its cloud capacity for AI.
Vendors and partners
Microsoft
Azure cloud foundation and Microsoft Foundry under Eliza; Microsoft 365 Copilot and GitHub Copilot · Sarthak Pattanaik said in Sep 2026 that BNY selected Azure first when it started building Eliza and then added Microsoft Foundry. Microsoft said in Oct 2026 that BNY worked with it through the Microsoft CEO Co-Innovation program to build an AI research capability on Foundry that investigates complex data issues and recommends next steps.
OpenAI2025
Model provider in Eliza under a multi-year collaboration · BNY's CEO Robin Vince joined the boards of the OpenAI Foundation and OpenAI Group PBC in Jul 2026, and BNY said in Apr 2026 that it uses an early-access OpenAI security capability.
Google CloudDec 2025
Gemini Enterprise integrated into Eliza; Gemini 3 and Veo 3 models · Google Cloud called BNY a long-time user of its AI capabilities in the Dec 8, 2025 announcement.
Anthropic
Model provider in Eliza; Mythos for cyber defense · BNY's 2025 annual report lists Anthropic among Eliza's model providers, and Robin Vince said in Apr 2026 that BNY runs Mythos in-house.
NVIDIA2024
DGX SuperPOD on-premises GPU infrastructure
Carnegie Mellon UniversitySep 2025
BNY AI Lab, a five-year, $10 million research agreement · Robin Vince said the BNY AI Lab opened at Carnegie Mellon in Sep 2026; the lab works on governance of mission-critical AI, verification of AI-generated code and evaluation of agents for critical systems.
DomynApr 2026
Financial AI company in which BNY invested · Domyn announced investments from BNY and Rabobank on Apr 20, 2026, and Carolyn Weinberg said BNY will explore how Domyn's technology can differentiate BNY's products.
Snorkel AIAug 2024
Software and expertise for turning unstructured data into AI models and insights · Snorkel AI announced a strategic agreement with BNY on Aug 8, 2024, when Sarthak Pattanaik headed BNY's AI Hub.
Kinexys by J.P. Morgan
Project Aikya, a federated learning proof of concept for anomaly detection in payments
MIT CSAIL; CRAFT at Stevens Institute of Technology and Rensselaer Polytechnic Institute
Academic AI research collaborations
Leadership
AI, data and platform leaders
Executives
Board
AI org map
Central AI Hub and one enterprise platform, Eliza, with AI built across BNY's business platforms
BNY created an AI Hub in 2023 to bring its data science, AI and machine learning teams together, launched Eliza in 2024, and named the AI Hub's head, Sarthak Pattanaik, Chief Data and AI Officer by Aug 2025. CIO and Global Head of Engineering Leigh-Ann Russell reports to CEO Robin Vince and sits on the Executive Committee; BNY has not published Pattanaik's reporting line. BNY also moved more than 70% of its employees into a platforms operating model by the end of 2025, and Robin Vince says that model, its commercial model and its culture give BNY an advantage in embedding AI.
BNY's AI work sits in three places: the AI Hub created in 2023, which brought BNY's data science, AI and machine learning teams together and whose former head, Sarthak Pattanaik, became Chief Data and AI Officer in 2025; the global engineering organization under CIO Leigh-Ann Russell; and BNY employees at large, more than half of whom were building AI agents in Apr 2026. BNY has not published how many AI specialists it employs; BNY said in Sep 2025 that it is the largest employer of graduates of Carnegie Mellon's Master of Science in Artificial Intelligence and Innovation program.
Reporting lines
Chairman and CEO
Engineering
Leigh-Ann Russell, CIO and Global Head of Engineering
Reporting line set in the Jun 2024 announcement; Executive Committee member
Finance
Dermot McDonogh, CFO
Operations and enterprise platforms
Alejandro Perez, COO
Executive Committee member
Product and Innovation
Carolyn Weinberg, Chief Product and Innovation Officer
Executive Committee member; Weinberg joined in Feb 2025
Chief Data and AI Office
Grew out of the AI Hub created in 2023; BNY has not published the office's reporting line
Board Technology Committee
M. Amy Gilliland, chair
Reviews technology strategy and receives regular reports on AI developments and governance
BNY's leadership page, checked Nov 1, 2026, lists 20 Executive Committee members and no one with a data or AI title; Leigh-Ann Russell, Carolyn Weinberg and Alejandro Perez are the technology, product and operations members. BNY has published a reporting line only for Russell, who reports to Robin Vince under the Jun 2024 announcement; Dermot McDonogh, Perez and Weinberg sit on Vince's Executive Committee, and BNY has not published who Sarthak Pattanaik reports to.
Where the AI workforce sits
AI Hub and Chief Data and AI Office
BNY's central data science, AI and machine learning teams, brought together in 2023; BNY has not published their size.
Not disclosedEngineering
BNY said AI authored more than 40% of its code in the first quarter of 2026; BNY has not published its engineering headcount.
Not disclosedEmployees building AI agents across the business
Over 50% of employees were building AI agents in Apr 2026 (nearly 50% in the Feb 2026 annual report); most AI builders come from outside engineering.
50%+ of staff
Named people
Use cases
BNY had more than 400 AI solutions in production in Oct 2026, Robin Vince said, alongside about 140 digital employees, spanning payments, custody, client onboarding, sales, legal and software engineering. BNY has published operating measures for a handful of these, such as faster onboarding and settlement inquiries and more software releases, which Robin Vince called inputs to AI's bottom-line impact, but no business-level return for any single use case. For AI as a whole, BNY says it is generating a return but does not publish AI spending, savings or revenue.
| Use case | Area | Status | ROI |
|---|---|---|---|
| Payment validation and repair BNY's digital employees validate payment details and repair missing or incomplete payment instructions so payments proceed without a time-consuming manual review; one digital employee handles more than 10% of BNY's payment repair issues worldwide (Sep 2026). | Payments | In production | |
| Restricted party screening AI reviews about 70% of BNY's restricted party screening for payments, and screening alerts reviewed by AI are resolved more than 30% faster (Apr 2026). | Payments | In production | |
| Multi-currency payments processing BNY embeds AI in high-volume multi-currency payments processing to automate it at scale. | Payments | In production | |
| Settlement inquiry resolution AI addressed more than 10% of BNY's custody settlement transaction status inquiries in the first quarter of 2026, and those inquiries were investigated more than 80% faster than a year earlier. | Asset Servicing | In production | |
| NAV anomaly detection BNY applies domain-trained AI to detect anomalies in net asset value calculations for funds it services. | Asset Servicing | In production | |
| Agentic onboarding research An agentic workflow reviews new client documents, draws on earlier documentation, gathers new data and supports the risk decision; about 25% of new onboardings used an AI-enabled process in the first quarter of 2026, and certain corporate onboarding processes finished more than 20% faster. | Client onboarding | In production | |
| Account planning and lead identification About 50% of any BNY annual account plan is drafted with AI as of 2025, and account plans were completed about 60% faster than in 2024. | Sales and client coverage | In production | |
| AI coding and digital engineers BNY engineers use GitHub Copilot through Eliza and work with digital engineers. AI authored more than 40% of BNY's code in the first quarter of 2026, when software releases rose more than 10% from a year earlier, and Saed Shonnar put new code developed with AI assistance at close to 60% in Sep 2026. | Engineering | In production | |
| Code vulnerability remediation BNY's digital employees remediate code vulnerabilities alongside engineering teams, Robin Vince said in Jan 2026. | Engineering | In production | |
| Contract review and contract lifecycle management BNY's legal department integrated Eliza to analyze executed contracts in its Universal Contract Master in 2025, and an AI Contract Review Assistant benchmarks negotiated agreements against corporate best practices and regulatory requirements. | Legal | In production | |
| Reconciliation automation BNY uses AI solutions to raise the share of reconciliations that run automatically (Oct 2025). | Operations | In production | |
| Transaction and market risk monitoring BNY uses AI to monitor transactions and market conditions continuously and flag risk signals closer to real time (Oct 2025). | Risk | In production | |
| Deep research with Gemini Enterprise BNY employees use Gemini Enterprise inside Eliza for agentic deep research on market analysis and build agents that synthesize financial reports, data and historical trends. | Markets and research | In production | |
| Portfolio-level credit risk analysis for clients BNY delivers client-specific AI solutions, including portfolio-level credit risk analysis. | Credit | In production | |
| Cyber defense with frontier models BNY runs Anthropic's Mythos model in-house and an early-access OpenAI security capability as part of its cyber defense, alongside AI and machine learning monitoring in its Cyber, Technology and Operations Center. | Cybersecurity | Pilot | |
| Project Aikya federated learning Project Aikya, a proof of concept with Kinexys by J.P. Morgan, tests federated learning to detect anomalies in cross-border payments without pooling the banks' data. | Payments | Pilot | |
| AI workflow capabilities for clients Robin Vince said in Oct 2026 that BNY sees an opportunity over time to extend Eliza's process and workflow capabilities to clients. | Client solutions | Announced |
Adoption
BNY spread AI by giving every employee the same platform and training them on it. Eliza adoption rose 175% in 2025, and the share of employees fully trained and onboarded went from just over a third in Jan 2025 to 99% by Oct 2025; by Apr 2026 every employee had access, about 50% used AI daily and over 50% were building AI agents. BNY delivered 171,000 AI learning hours in 2025, and Microsoft counted more than 400,000 hours of BNY AI training by Sep 2026, when CIO Leigh-Ann Russell said almost two-thirds of employees were daily or super users of AI and almost 60% had been onboarded to Microsoft 365 Copilot. BNY's count of AI solutions in production went from 40 in 2024 to 160 at the end of 2025, about 220 in Apr 2026 and more than 400 in Oct 2026.
By the numbers
Rollout, first pilot to today
2023
BNY created its AI Hub, bringing its data science, AI and machine learning teams together, and identified 30 areas of opportunity for AI.
2024
BNY became the first global bank to deploy an NVIDIA DGX SuperPOD and launched Eliza, its enterprise AI platform, the same year.
Sep 2024
Leigh-Ann Russell joined BNY from bp as Chief Information Officer and Global Head of Engineering, reporting to Robin Vince.
Aug 2025
CDO Magazine reported that BNY named Sarthak Pattanaik, head of its AI Hub, Chief Data and AI Officer.
Sep 2025
BNY and Carnegie Mellon University agreed a five-year, $10 million collaboration to set up the BNY AI Lab.
Sep 2025
Robin Vince unveiled a next-generation Eliza at an internal town hall, letting employees build their own agents and draw on data from across BNY and live market feeds.
Oct 2025
BNY said Eliza adoption was up 175% in 2025, 99% of employees were trained and onboarded, AI solutions in production rose 75% in a quarter, and more than 100 digital employees were deployed.
Dec 2025
BNY integrated Google Cloud's Gemini Enterprise into Eliza for deep research and multimodal tools.
Jan 2026
BNY reported that it launched Eliza 2.0 in 2025 and ended the year with 99% of employees on Eliza, 160 enterprise AI solutions in production and 134 digital employees.
Apr 2026
BNY reported about 220 enterprise AI solutions in production, about 140 digital employees, Eliza access for 100% of employees and about 50% of employees using AI daily.
Sep 2026
Microsoft reported more than 300 AI use cases in production at BNY, 400,000 hours of AI training, and almost 60% of BNY onboarded to Microsoft 365 Copilot.
Oct 2026
Robin Vince said BNY has more than 400 AI solutions in production.
Economics & controls
Spend and ROI
Technology budget
Robin Vince said on Oct 15, 2026 that BNY spends $4 billion a year on technology within a $13 billion expense base, and Dermot McDonogh described an engineering budget of about $4 billion in Jul 2026. BNY's software and equipment expense was $2.15 billion in 2025, up 9%. BNY generated about $550 million of efficiency savings and about $500 million of incremental investment in new client solutions, technology and people in 2025, and said in Jul 2026 that it is tracking toward about $450 million of savings and $500 million of incremental investment in 2026; BNY does not say how much of either comes from AI.
What AI itself costs
BNY does not break out its AI spending. Dermot McDonogh said in Jan 2026 that BNY spends more on cyber resiliency than on AI, and called AI spending quite modest within the roughly $4 billion engineering budget in Jul 2026, when he told Fortune that BNY focuses on outcomes rather than counting tokens. Robin Vince said in Oct 2026 that AI spending folds into BNY's expense base, that breaking it out could be counterproductive, and that BNY measures token costs precisely and routes work to cheaper models inside Eliza. McDonogh said on Oct 27, 2026 that BNY built telemetry to track how it spends AI tokens.
Reported ROI
BNY says AI is generating a return but publishes no dollar figure and no business-level return for any AI use case. Robin Vince said in Jul 2026 that BNY can identify AI benefits but treats them as a package aimed at growing revenue faster than expenses, and called the AI statistics BNY published in Apr 2026 inputs to AI's eventual bottom-line impact. Vince said in Oct 2026 that the answer to whether AI is earning a return is yes.
Productivity markers
BNY's Apr 2026 investor presentation listed AI outputs across the client lifecycle: account plans completed about 60% faster, certain corporate onboarding processes more than 20% faster, restricted party screening resolved more than 30% faster when reviewed by AI, settlement inquiries investigated more than 80% faster, and software releases up more than 10% as AI authored more than 40% of code. BNY's revenue per employee, excluding notable items, rose from $338,000 in 2022 to $401,000 in 2025; BNY says the gains will increasingly be enabled by AI but does not credit them to AI. Robin Vince said in Apr 2026 that the progress so far came largely from the platforms operating model and commercial model, with AI powering the next leg, and Dermot McDonogh made the same point in Jul 2026.
Headcount
BNY had 46,600 full-time employees on Sep 30, 2026, down from 49,200 a year earlier and from 53,400 at the end of 2023. Dermot McDonogh said in Jan 2026 that the decline was not yet related to AI and in Jul 2026 called headcount an output, not a target, noting that BNY's early careers class is three times larger than three years earlier. BNY's 2025 Form 10-K says AI-driven automation may change the composition of its workforce and require investment in reskilling.
Governance and risk
Governance built into Eliza
BNY built governance into Eliza from the start, and CIO Leigh-Ann Russell calls Eliza BNY's AI governance dashboard as well as its technology harness (Sep 2026). BNY's Responsible AI approach covers data usage, transparency, fairness, compliance, employee training and technical guardrails (Oct 2025), new AI ideas go to a review committee that meets daily (Sep 2026), and BNY's AI governance framework draws on its legal, privacy, responsible AI, data governance, information security, resiliency, and risk and compliance teams.
Controls on digital employees
BNY gives each digital employee its own persona, credentials and a human supervisor and deploys agents with guardrails, telemetry and approvals (Oct 2025). Robin Vince said in Oct 2026 that BNY does not make payments without extensive validation and that AI can run inside the same maker-checker and challenger processes as staff. BNY's 2025 Form 10-K lists digital employees and digital engineers as a source of operational and control risk, including errors in automated decision-making and challenges in oversight and accountability.
Model risk and data
BNY's Model Risk Management function, which is independent of model developers and users, covers use cases that employ AI, including large language models, and sits inside a cross-functional AI risk management framework, according to the 2025 annual report. BNY's responsible AI commitment draws on the NIST AI Risk Management Framework and the Monetary Authority of Singapore's Veritas consortium and requires human oversight and explainable decisions. BNY's 10-K warns that BNY may be unable to control how third-party AI models are trained or how data it inputs is retained or reused.
Board oversight and pay
BNY's board Technology Committee, chaired by M. Amy Gilliland since Jul 1, 2025, receives regular reports on AI developments and governance, and directors receive education sessions on AI. BNY's 2025 executive pay scorecard counted the Eliza 2.0 launch, 160 AI solutions and 134 digital employees among its results, according to the 2026 proxy statement.
Regulation
BNY has set up an implementation program for the EU AI Act, which it says will apply to BNY, and is subject to New York Department of Financial Services rules that require AI risks to be addressed in its cybersecurity program, according to the 2025 annual report.
Cyber defense and model safety
Robin Vince said in Apr 2026 that BNY runs Anthropic's Mythos model in-house as part of its cyber defense, and in Oct 2026 that AI alignment and observability research must catch up with frontier models and that open-weight models carry risk because users can remove safety features. The BNY AI Lab at Carnegie Mellon, opened in Sep 2026, researches verification of AI-generated code and evaluation of agents for critical systems.
Recent signals
Oct 27, 2026
Dermot McDonogh told Bloomberg Television that BNY built telemetry to track how it spends AI tokens and routes simple queries away from expensive models.
Oct 27, 2026
Microsoft said more than 300 solutions run on Eliza and that BNY worked with Microsoft through its CEO Co-Innovation program to build an AI research capability on Foundry that traces the root causes of complex data issues.
Oct 15, 2026
Robin Vince said BNY has more than 400 AI solutions in production, spends $4 billion a year on technology within a $13 billion expense base, does not break out AI spending, and uses open-source models with routing in Eliza to manage token costs.
Oct 1, 2026
American Banker reported that BNY's digital employees, developed under CIO Leigh-Ann Russell, number more than 150 and that over 60% of BNY's code is now built with AI's help.
Sep 24, 2026
Microsoft published BNY case studies reporting more than 300 AI use cases in production, almost two-thirds of employees as daily or super users of AI and almost 60% onboarded to Microsoft 365 Copilot.
Sep 2026
BNY opened the BNY AI Lab at Carnegie Mellon University, researching verification of AI-generated code and evaluation of agents for critical systems, Robin Vince said in Oct 2026.
Jul 28, 2026
Dermot McDonogh told Fortune that BNY tracks AI outcomes rather than token counts, called token costs modest, and named regulatory reporting, balance sheet analytics and predictive modeling as finance use cases.
Jul 21, 2026
OpenAI named BNY CEO Robin Vince to the boards of the OpenAI Foundation and OpenAI Group PBC.
Jul 15, 2026
Dermot McDonogh called BNY's AI spending quite modest within an engineering budget of about $4 billion and said headcount, down 7% in a year, is an output, while Robin Vince said AI is earning a return BNY does not break out.
Apr 20, 2026
Domyn, a financial AI company, announced investments from BNY and Rabobank; Carolyn Weinberg said BNY will explore Domyn's technology for its products.
Apr 16, 2026
BNY reported about 220 AI solutions in production, about 140 digital employees and AI-driven gains in onboarding, screening, settlement inquiries and coding, and Robin Vince said BNY runs Anthropic's Mythos model in-house for cyber defense.
Mar 5, 2026
BNY's proxy statement said the board's Technology Committee receives regular AI reports and that the 2025 executive pay scorecard counted the Eliza 2.0 launch and 134 digital employees.
Feb 25, 2026
BNY's 2025 annual report said Eliza integrates OpenAI, Google and Anthropic models, 160 AI solutions and 134 digital employees were in production, and 171,000 AI learning hours were delivered in 2025.
Jan 13, 2026
BNY reported the Eliza 2.0 launch and 134 digital employees for 2025, and Dermot McDonogh said the decline in headcount was not yet due to AI and that BNY spends more on cyber resiliency than on AI.
Dec 8, 2025
BNY and Google Cloud integrated Gemini Enterprise into Eliza for deep research on market analysis, using Gemini 3 and Veo 3.
Quick answers
- How does BNY deploy AI?
- BNY deploys AI through Eliza, an enterprise platform it launched in 2024 on Microsoft Azure that routes work across OpenAI, Google, Anthropic and open-source models, backed by its own NVIDIA DGX SuperPOD. Every BNY employee has Eliza access, about half used AI daily in Apr 2026, and CEO Robin Vince said in Oct 2026 that more than 400 AI solutions run in production, alongside digital employees that repair payments and fix code vulnerabilities. BNY says AI earns a return but does not publish AI spending or savings.
- What does BNY use AI for?
- BNY had more than 400 AI solutions in production in Oct 2026, Robin Vince said, alongside about 140 digital employees, spanning payments, custody, client onboarding, sales, legal and software engineering. BNY has published operating measures for a handful of these, such as faster onboarding and settlement inquiries and more software releases, which Robin Vince called inputs to AI's bottom-line impact, but no business-level return for any single use case. For AI as a whole, BNY says it is generating a return but does not publish AI spending, savings or revenue.
- Who leads AI at BNY?
- AI at BNY is led by Leigh-Ann Russell (Chief Information Officer and Global Head of Engineering); Sarthak Pattanaik (Chief Data and AI Officer); Saed Shonnar (Head of AI Enablement). Ownership of AI on the org chart: Chief Data and AI Officer Sarthak Pattanaik; CIO Leigh-Ann Russell reports to the CEO.
- What AI platform does BNY use?
- Eliza, the enterprise AI platform BNY launched in 2024 and moved to Eliza 2.0 in 2025, puts chat, deep research, agent building and AI governance in one place and runs on Microsoft Azure and Microsoft Foundry (Sep 2026).
- Which AI models does BNY use?
- Eliza is model-agnostic: it integrates OpenAI, Google and Anthropic models (2025 annual report), and Robin Vince said in Oct 2026 that BNY also uses open-source models and routes each task to the model that fits; GitHub Copilot and Microsoft 365 Copilot sit inside Eliza.
- How many BNY employees use AI?
- Every BNY employee had Eliza access and about 50% used AI daily in Apr 2026; CIO Leigh-Ann Russell put daily or super users at almost two-thirds of staff in Sep 2026.
- How far has AI spread at BNY?
- BNY has reported no business return it credits to AI, for any use case or for the firm as a whole: Robin Vince said on Oct 15, 2026 that AI is generating a return, without giving a figure, and that BNY does not break AI out of its expense base; the AI gains BNY publishes are operating measures, such as settlement inquiries investigated more than 80% faster; and BNY does not say how much of its roughly $550 million of 2025 efficiency savings came from AI. BNY has officially reported a company-wide rollout, which places it in Piloting: its Apr 2026 investor presentation said every employee has access to Eliza and about 50% use AI daily, and Chief Data and AI Officer Sarthak Pattanaik owns the program. BNY falls short of Scaling because it reports no AI-credited return in any function and no firm-level AI figure, and it presents its higher revenue per employee as the result of reimagining the company, not as a return from AI.
Open questions
- How much does BNY spend on AI, including tokens and GPUs, given that Robin Vince declines to break the figure out of the $13 billion expense base and Dermot McDonogh says BNY now tracks token spending with its own telemetry (Oct 2026)?
- What dollar savings or revenue have BNY's digital employees and more than 400 AI solutions produced, and how much of the roughly $550 million of 2025 efficiency savings came from AI?
- What business return do BNY's deployed AI tools in payments, onboarding, settlement inquiries, contract review and software engineering deliver, beyond the faster processing times and higher release counts BNY has published, and how large are the cost savings the 2026 proxy credits to the Eliza-based contract program?
- How many AI and machine learning specialists does BNY employ, and how large is the AI Hub?
- Who does Chief Data and AI Officer Sarthak Pattanaik report to?
- Where do the OpenAI and Anthropic models in Eliza run, through Microsoft Foundry on Azure, directly with the labs or on BNY's own NVIDIA hardware, and which workloads run on the DGX SuperPOD?
- How does BNY count AI solutions and use cases, given that Microsoft reported more than 300 use cases in Sep 2026 and Robin Vince more than 400 solutions in Oct 2026?
- How much of the fall in full-time employees from 53,400 at the end of 2023 to 46,600 in Sep 2026 is due to AI, given Dermot McDonogh's Jan 2026 statement that the decline was not yet related to AI?
- How does BNY manage potential conflicts between Robin Vince's seats on OpenAI's boards and BNY's choice of model providers?
- When will BNY offer Eliza-based workflow capabilities to clients, what did Project Aikya and the Mythos testing find, and will Domyn's models run inside Eliza?
Sources
BNY
- BNY: Artificial intelligence at BNY
- BNY Insights: AI in the Next Phase: Driving Adoption (Oct 20, 2025)
- BNY: Commitment to the responsible, ethical use of data and AI
- BNY: Project Aikya with Kinexys by J.P. Morgan
- BNY and Carnegie Mellon University join forces on AI research (Sep 11, 2025)
- Leigh-Ann Russell to join BNY as Chief Information Officer (Jun 2024)
- Carolyn Weinberg joins BNY as Chief Solutions Innovation Officer (Jan 30, 2025)
- BNY leadership and Executive Committee
- Robin Vince, CEO, biography
- Leigh-Ann Russell, CIO and Global Head of Engineering, biography
- Carolyn Weinberg, Chief Product and Innovation Officer, biography
- Alejandro Perez, COO, biography
- Dermot McDonogh, CFO, biography
Filings, letters and investor materials
- Annual Report 2025: Built for Opportunity (Feb 2026)
- Annual Report 2025, full PDF with Robin Vince's letter
- 2025 Annual Report, Exhibit 13.1 to the Form 10-K (Feb 2026)
- BNY 2025 Form 10-K (filed Feb 2026)
- BNY 2026 proxy statement (Mar 5, 2026)
- 4Q25 quarterly update presentation (Jan 13, 2026)
- 1Q26 quarterly update presentation (Apr 16, 2026)
- 2Q26 quarterly update presentation (Jul 15, 2026)
- 3Q26 financial supplement (Oct 15, 2026)
- Q3 2026 earnings call transcript, The Motley Fool via Webull (Oct 15, 2026)
- Q2 2026 earnings call transcript, Quartr via Fortune (Jul 15, 2026)
- Q1 2026 earnings call transcript, Quartr via Fortune (Apr 16, 2026)
- Q4 2025 earnings call transcript, Quartr via Fortune (Jan 13, 2026)
Partners
- Google Cloud: BNY collaborates with Google Cloud to advance Eliza with Gemini Enterprise (Dec 8, 2025)
- Microsoft: BNY upskills 100% of staff on Eliza, powered by Microsoft AI (Sep 24, 2026)
- Microsoft: BNY resolves client inquiries 80% faster with Eliza (Sep 24, 2026)
- Microsoft: The platform approach to becoming Frontier, by Judson Althoff (Oct 27, 2026)
- OpenAI Foundation: David Vélez and Robin Vince join the boards (Jul 21, 2026)
- Domyn: Domyn secures BNY and Rabobank investments (Apr 20, 2026)
- Snorkel AI signs agreement with BNY, via Stock Titan (Aug 8, 2024)
News
- Bloomberg Law: OpenAI names BNY, Nubank CEOs to board ahead of IPO (Jul 21, 2026)
- The Next Web: BNY says it now tracks its AI token spending (Oct 27, 2026)
- American Banker: While embedding AI everywhere, BNY won't disclose costs (Oct 15, 2026)
- American Banker: Most Powerful Women in Banking, Leigh-Ann Russell (Oct 1, 2026)
- Fortune: BNY skipped the tokenmaxxing craze (Jul 28, 2026)
- American Banker: BNY says AI bet is paying off, client impact is measurable (Jul 15, 2026)
- Fortune: BNY unveils updated Eliza platform (Sep 29, 2025)
- CDO Magazine: BNY names Sarthak Pattanaik as Chief Data and AI Officer (Aug 2025)
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
BNY profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 1, 2026
