Verified
Bank of America
Scaling- Banking
- 209,911 employees
- $14B tech spend
- Erica, built in-house
- ~200,000 active AI users
- Evident AI Index #7 (2026)
Bank of America deploys AI through Erica, the virtual assistant it launched in 2018 and reuses for clients, advisers and staff, plus general AI tools that more than 200,000 employees were using in Oct 2026 and role-specific tools for advisers, bankers, call center staff and developers. Bank of America had more than 300 approved AI use cases in Oct 2026. CTO and CIO Hari Gopalkrishnan builds the technology, and strategy head Jeff Busconi runs the AI catalyst program that pushes usage and value.
about.bankofamerica.comTheir approach to AI
- The platform
- Erica, which Bank of America built in-house and launched in 2018, now also powers Erica for Employees, EricaAssist, ask MERRILL and CashPro Chat, alongside in-house generative AI tools such as AskGPS and Orchestra.
- The models
- Bank of America gives staff multiple open-source and proprietary models in cloud environments it does not name, uses AI built into vendor tools such as Microsoft Copilot, GitHub Copilot and Salesforce Agentforce, and keeps customer data out of outside models (Apr 2026).
- The reach
- About 200,000 Bank of America employees were active users of AI tools in Jul 2026, sending more than 400,000 prompts a day, and 24.2 million clients actively used Erica in the third quarter of 2026.
- The catch
- Bank of America's firm-level AI return rests on CEO remarks: Brian Moynihan said in Sep 2026 that 130 to 140 projects costing $400 million generate $800 million of benefit, still being captured, and in Oct 2026 that none is a huge contributor.
Snapshot
- Employees
- 209,911
- Sep 30, 2026, down from 213,384 a year earlier (3Q26 earnings release)
- Technology budget
- $14B a year
- More than $4B on new initiatives including AI (Jul 2026); $13B spent in 2025
- AI/ML headcount
- Not disclosed
- About 19,000 developers use AI coding assistance (Jul 2026); 8,200+ inventors (Nov 2025)
- Who owns AI
- Hari Gopalkrishnan, CTO and CIO
- Jeff Busconi drives AI usage and value as AI catalyst; Thong Nguyen oversees AI governance
- Platform
- Erica, built in-house
- Launched 2018; reused for employees, advisers and CashPro clients
- Models
- Open-source and proprietary
- Providers not named; Microsoft Copilot and GitHub Copilot in use (Nov 2025)
- Reach
- ~200,000 active AI users
- Jul 2026, sending more than 400,000 prompts a day
- Use cases
- 300+ approved
- Oct 2026; 114 live generative AI use cases, 34 fully deployed (Jul 2026)
Key findings
About 200,000 Bank of America employees were active users of AI tools in Jul 2026, sending more than 400,000 prompts a day, after the 2025 annual report in Mar 2026 counted about 150,000 active users and more than 1.5 million prompts a week.
Bank of America had more than 300 approved AI and machine learning use cases in Jul 2026, 114 of them live generative AI use cases and 34 fully deployed, and Brian Moynihan said on Sep 14, 2026 that 130 to 140 implemented AI projects cost $400 million and generate $800 million of benefit, which the bank is still capturing.
Bank of America spends about $14 billion a year on technology, more than $4 billion of it on new initiatives including AI (Jul 2026), and does not break out what AI itself costs.
Erica, the virtual assistant Bank of America launched in 2018, had 24.2 million active users in the third quarter of 2026, up 18% on a year earlier, and Erica for Employees has cut calls to the IT service desk by more than 55% since 2020.
Bank of America's headcount fell to 209,911 at Sep 30, 2026 from 213,384 a year earlier as it automated processes, and Brian Moynihan said managers fell about 5% while client-facing staff grew.
Where BofA sits
- Stage
- Scaling
Bank of America sits at Scaling: all of its roughly 209,000 employees have access to AI tools, more than 200,000 were using them in Oct 2026, more than 300 AI use cases were approved by Oct 2026, and Jeff Busconi is accountable for AI usage and value as the company's AI catalyst. Bank of America falls short of Embedded on the financial test: its only firm-level return is Brian Moynihan's Sep 2026 conference estimate that 130 to 140 projects costing $400 million generate $800 million of benefit still being captured, and he said in Oct 2026 that none of the projects is a huge contributor. Usage is close to the Embedded bar, but Bank of America has not defined an active user or said how often employees use AI.
How the stages are defined - Ownership
- CTO and CIO Hari Gopalkrishnan builds; Jeff Busconi drives usage as AI catalyst
- Platform
- Hybrid: Erica and generative AI tools built in-house; Copilot and Agentforce bought
- Models
- Open-source and proprietary models, providers not named; AI in vendor tools
- Hosting
- Cloud hosting environments the bank does not name; customer data kept out of outside models
- Reach
- About 200,000 active AI users (Jul 2026)
How they deploy
Build once, reuse across businesses
Bank of America built Erica in-house and launched it for consumers in 2018, then reused the same technology for Erica for Employees (2020), ask MERRILL and ask PRIVATE BANK, CashPro Chat for business clients (Aug 2023) and EricaAssist for call center staff. Bank of America said in Aug 2025 that its AI capabilities are designed to be scalable and reusable across different areas.
Three levels of AI tools
Bank of America's Jul 2026 earnings deck sorts its AI into three levels: general-purpose productivity tools built into enterprise applications, used by about 200,000 people; functional tools built for specific roles and tied to core systems, such as meeting tools for more than 20,000 wealth professionals and coding help for about 19,000 developers; and custom multi-process systems such as call center recommendations and an enterprise document intelligence platform. All three sit on access to multiple open-source and proprietary models.
An employee-led AI catalyst program
Brian Moynihan wrote in Mar 2026 that Bank of America had launched an AI catalyst program a few months earlier, under a corporate leader with champions across the company, to gather ideas from staff and outside experts and turn them into projects with business cases. Jeff Busconi, head of corporate strategy and Operational Excellence, serves as the AI catalyst with overall responsibility for AI innovation, usage and value, and Moynihan said in May 2026 that 22 colleagues led by Busconi work on the program.
Humans stay in the loop
Bank of America describes its approach as high-tech, high-touch, with human oversight, transparency and accountability for all AI outcomes (Apr 2025). Brian Moynihan said in Oct 2026 that the bank is working with AI agent providers and aggregators so that a human approves before money moves out of an account.
More output from a steady technology budget
Brian Moynihan said in Jul 2026 that AI coding tools mean the same technology spending in 2027 will produce more code in 2028, and in Oct 2026 that Bank of America's technology development spending of a little over $4 billion a year is 10% more efficient, so the bank gets 10% more work for the same cost.
- Deployment model
- Bank of America builds AI as enterprise platforms in Global Technology, led by Chief Technology and Information Officer Hari Gopalkrishnan, and reuses them across its eight lines of business. Jeff Busconi runs the AI catalyst program that pushes usage and value company-wide, Vice Chair Thong Nguyen oversees enterprise data and AI governance, and the 2026 proxy statement describes an internal AI council that governs generative AI tools. Co-Presidents Dean Athanasia and Jim DeMare took on the expansion of AI-based tools for clients in Sep 2025, and business lines run their own data and AI teams, such as Jarrett Bruhn's in Global Payments Solutions.
- Model providers
- Bank of America has not named the companies whose models power its generative AI tools. Bank of America's Jul 2026 AI slide says employees have access to multiple open-source and proprietary models in enabled cloud hosting environments. The investment bank's generative AI suite includes Microsoft Copilot, which logged more than 250,000 prompts in its first month, and about 18,000 software developers use GitHub Copilot (Nov 2025 investor day). Brian Moynihan said in May 2026 that 4,000 to 5,000 employees were using Salesforce's Agentforce, and in Sep 2026 that the general AI providers embed AI in the platforms they supply to the bank and that Bank of America has access to specialized models where needed. Erica was trained by Bank of America's own data scientists on a library of more than 700 responses and has had more than 75,000 updates since launch (Aug 2025).
- Data and infrastructure
- Bank of America has invested more than $1.5 billion in data capabilities since 2021, holds more than 918 petabytes of storage and runs about 940,000 grid CPU cores (second quarter of 2025, Nov 2025 investor day). Bank of America maps about 3,700 processes and 55,000 activities in a single process inventory that it uses to find work for AI and automation. Brian Moynihan said in Apr 2026 that Bank of America keeps customer data out of the models, taking the models in without feeding them, and that protecting customer data has slowed its adoption of some technologies. Bank of America has not named the cloud providers or data centers that host its models.
Platforms and internal tools
EricaJun 2018
Erica is Bank of America's in-house AI virtual financial assistant in its consumer mobile app, which answers client questions and sends proactive, personalized insights. Clients used Erica nearly 700 million times in 2025 and more than 3.2 billion times since launch, and more than 98% of users find what they need without a call or chat (Sep 2025).
- Users
- 24.2 million active users (third quarter of 2026)
Erica for Employees2020
Erica for Employees is Bank of America's internal virtual assistant for technology support, which expanded in 2023 to human resources questions such as health benefits and payroll forms.
- Users
- More than 90% of employees (Apr 2025)
EricaAssistBy Nov 2025
EricaAssist is a human-assisted AI agent that works alongside Bank of America customer service representatives during calls, summarizing why a client is calling and recommending next steps. Bank of America says EricaAssist cuts average call time by nearly one minute, and generative AI features added in Jul 2026 return guidance in under three seconds.
- Users
- More than 18,000 customer service representatives (Jul 2026)
ask MERRILL and ask PRIVATE BANK
ask MERRILL and ask PRIVATE BANK use Erica's technology to help Merrill and Private Bank teams find and curate information and reach experts inside Bank of America.
- Users
- About 23 million interactions a year (Mar 2026)
CashPro Chat with EricaAug 2023
CashPro Chat is the virtual service advisor on Bank of America's CashPro treasury platform, built with Erica technology, which handles more than 40% of business client inquiries on its own.
- Users
- 65% of business clients (Aug 2025)
AskGPS and AskGPS Intelligence HubSep 2025
AskGPS is a generative AI assistant Bank of America built in-house for its Global Payments Solutions staff, trained on more than 3,200 internal documents; the AskGPS Intelligence Hub, announced in Sep 2026, adds treasury reviews, account schematics and relationship insights drawn from client data.
- Users
- Nearly 3,000 employees (Sep 2026)
BofA Orchestra AIBy 2025
Orchestra is a proprietary generative AI chatbot connected to Bank of America research and market commentary, used by Global Markets and Global Corporate and Investment Banking to search, summarize and synthesize research and outside financial content.
AI-Powered Meeting JourneyMar 2026
AI-Powered Meeting Journey prepares meeting materials, takes notes with client consent and drafts follow-up tasks for Merrill and Bank of America Private Bank advisers, rolled out at full scale in Mar 2026.
- Users
- More than 20,000 wealth professionals with AI meeting tools (Jul 2026)
Enterprise document intelligence platform
Bank of America's document intelligence platform searches, classifies, extracts and translates documents to speed client servicing and cut manual work, and includes an in-house generative AI solution that processes death certificates for estate servicing.
CashPro Data Intelligence
CashPro Data Intelligence is Bank of America's suite of AI-powered treasury insights and analytics tools for CashPro clients, including Payments Insights (added Sep 2026), Capital Markets Insights, Mobile Security Insights, CashPro Forecasting and CashPro Search.
Vendors and partners
MicrosoftBy Nov 2025
Microsoft Copilot in the investment bank's generative AI suite · Bank of America said Microsoft Copilot logged more than 250,000 prompts in its first month of use in Global Corporate and Investment Banking (Nov 2025 investor day).
GitHub (Microsoft)By Nov 2025
GitHub Copilot coding assistant · About 18,000 Bank of America software developers used GitHub Copilot, according to the Nov 2025 investor day.
SalesforceBy May 2026
Agentforce AI agents for Bank of America staff · Brian Moynihan said on May 27, 2026 that Bank of America is a big Salesforce user and that 4,000 to 5,000 people were using Agentforce to prepare for meetings, growing by thousands a month (Bernstein conference transcript).
ASB Bank, Capital One, Commonwealth Bank of Australia, ING and NatWestSep 2026
Co-authors of joint principles for agentic commerce · Bank of America and five other banks published principles on transparency, safety, privacy and data, choice and interoperability for payments made by AI agents.
Leadership
AI, data and platform leaders
Executives
Board
AI org map
Enterprise AI platforms built in Global Technology; an AI catalyst for usage and value; data and AI teams in the business lines
Bank of America has no single chief AI officer. Chief Technology and Information Officer Hari Gopalkrishnan leads the technology organization that builds Erica and the bank's generative AI tools, Jeff Busconi drives AI usage and value as the company's AI catalyst, and Vice Chair Thong Nguyen oversees enterprise data and AI governance. Co-Presidents Dean Athanasia and Jim DeMare, appointed in Sep 2025, oversee the eight lines of business and the expansion of AI-based tools for clients, and the 2026 proxy statement credits an internal AI council with governing the rollout of generative AI tools.
Bank of America has not disclosed how many AI and machine learning specialists it employs. Bank of America's technology workforce includes about 19,000 developers who use AI coding assistance (Jul 2026), and the bank counted more than 8,200 inventors and about 1,500 AI and machine learning patents granted and pending in Nov 2025. Erica is trained by Bank of America's own data scientists, and business lines such as Global Payments Solutions have their own data and AI heads.
Reporting lines
Chair and CEO
Brian Moynihan
Co-Presidents
Dean Athanasia and Jim DeMare
Oversee the eight lines of business and the expansion of AI-based tools for clients (Sep 2025)
Marketing, Digital and Specialized Consumer Client Solutions
Dean Athanasia's biography lists Marketing and Digital among his responsibilities
Global Payments Solutions
Mark Monaco
Global Technology
Hari Gopalkrishnan, CTO and CIO
Member of the executive management team; the direct line to the CEO is inferred, not published
Consumer Technology
Tom Ellis, CIO
Builds Erica and EricaAssist for the consumer business; reporting line inferred from the CIO title
Global Strategy & Enterprise Platforms
Thong Nguyen, Vice Chair
Enterprise data, AI governance, payments strategy, business continuity and procurement; line to the CEO inferred
Corporate Strategy, Operational Excellence and AI catalyst
Jeff Busconi
Inferred from Thong Nguyen's biography, which lists corporate strategy and Operational Excellence under him; not published
Global Operations
Tom Scrivener
Reporting line not published
Bank of America does not publish reporting lines below the executive management team. Hari Gopalkrishnan, Thong Nguyen, Jeff Busconi, David Tyrie and Tom Scrivener all appear as members of the executive management team in their newsroom biographies, but the bank does not say whom each reports to. The 2026 proxy statement mentions an internal AI council without naming its chair or members.
Where the AI workforce sits
Wealth management professionals with AI meeting tools
Estimate of the share of role-specific AI tool users, computed from Bank of America's Jul 2026 figures (more than 20,000 of about 50,000); these are users, not AI staff.
~40%Developers with AI coding assistance
About 19,000 developers (Jul 2026), on the same basis.
~38%Call center agents with automated recommendations
About 5,000 agents (Jul 2026); EricaAssist alone counts more than 18,000 representatives, which would raise this share if the groups differ.
~10%Corporate and investment bankers using AI for research and preparation
About 4,000 professionals (Jul 2026).
~8%Global Payments Solutions sales and service staff
About 2,500 representatives (Jul 2026).
~5%
Named people
Use cases
Bank of America had more than 300 approved AI and machine learning use cases in Oct 2026, with 114 live generative AI use cases and 34 fully deployed in Jul 2026, across consumer service, wealth management, banking, payments, markets, operations and software development. Bank of America has published a business return for two of them, AI fraud detection and AI coding assistance. For AI as a whole, Brian Moynihan said in Sep 2026 that 130 to 140 implemented projects cost $400 million and generate $800 million of benefit still being captured, and in Oct 2026 that the projects return two to three times their cost over time, with none of them a huge contributor.
| Use case | Area | Status | ROI |
|---|---|---|---|
| Erica virtual assistant Erica answers consumer questions in the Bank of America app and sends proactive insights on spending, balances, rewards and credit, and hands clients to people when needed. | Consumer Banking | In production | |
| EricaAssist for customer service EricaAssist summarizes why a client is calling, pulls together relevant information and recommends next steps for more than 18,000 Bank of America customer service representatives. | Consumer Banking | In production | |
| AI fraud detection Bank of America runs more than 50 AI-enabled fraud detection models to spot fraudulent activity faster and at scale. | Consumer Banking | In production | Consumer fraud loss rate ~50% below 2018, credited to AI-enabled processes (Nov 2025)Source |
| AI-Powered Meeting Journey Merrill and Private Bank advisers use AI to prepare meeting materials, take notes with client consent and draft follow-up tasks, so they can serve more clients. | Global Wealth & Investment Management | In production | |
| ask MERRILL and ask PRIVATE BANK Wealth management teams use Erica-based tools to find resources, curate information for clients and reach specialists inside Bank of America. | Global Wealth & Investment Management | In production | |
| Merrill SENSE SENSE is an internally developed search engine built on a large language model that reads advisers' activity notes to match clients and prospects with Merrill events. | Global Wealth & Investment Management | In production | |
| Banker research and meeting preparation About 4,000 corporate and investment banking professionals use AI for research and presentation preparation, and Business Banking and Commercial Banking teams use it to draft client meeting materials. | Global Banking | In production | |
| Generative AI for commercial prospecting and onboarding Bank of America's Commercial Bank launched a generative AI tool in Sep 2026 to make prospecting and client onboarding more efficient. | Global Banking | In production | |
| Investment banking generative AI suite Global Corporate and Investment Banking pairs Orchestra, Microsoft Copilot, pitchbook page generators and bots for company overviews and working group lists. | Global Banking | In production | |
| CashPro Chat with Erica Erica answers business clients' questions on CashPro about payments, transaction status and account access, routing the rest to live experts. | Global Payments Solutions | In production | |
| AskGPS Nearly 3,000 Global Payments Solutions employees use AskGPS to answer client product and onboarding questions from vetted internal documents, and the Intelligence Hub adds AI analysis of client and account data. | Global Payments Solutions | In production | |
| CashPro Forecasting CashPro Forecasting applies machine learning to corporate clients' account data to forecast cash positions up to a year ahead, and Bank of America says it saved more than 3,000 client companies over 250,000 hours in 2025. | Global Payments Solutions | In production | |
| Orchestra research search and summaries Orchestra lets sales and trading staff search, summarize and synthesize Bank of America research and market commentary, and Global Markets also uses AI for automated daily news summaries and trade lifecycle work. | Global Markets | In production | |
| AI coding assistance About 19,000 Bank of America developers use real-time coding assistance, including GitHub Copilot, to write and optimize code. | Global Technology | In production | Tech development ($4B+ a year) 10% more efficient with AI: more work for the same cost (Oct 2026)Source |
| Erica for Employees Erica for Employees answers Bank of America employees' technology support and human resources questions, and Bank of America says it has cut IT service desk calls by more than 55% since its 2020 launch (Mar 2026). | Firmwide | In production | |
| General-purpose AI productivity tools Bank of America builds AI into enterprise-wide productivity applications, with more than 2,000 departmental AI templates in use (Jul 2026). | Firmwide | In production | |
| Enterprise document intelligence Global Operations uses AI to search, classify, extract and translate documents, including converting estate servicing paperwork such as death certificates into usable data. | Operations | In production | |
| Legal research and drafting Bank of America's legal teams use AI to support research, analysis and the preparation of contracts and documents. | Legal | In production | |
| Conversation simulators in The Academy The Academy, Bank of America's training organization, uses AI conversation simulators so employees can practice client conversations with real-time feedback. | Learning | In production |
Adoption
Bank of America spread AI in layers: Erica for consumers from Jun 2018, Erica for Employees from 2020, Erica inside CashPro Chat from Aug 2023, then generative AI tools for staff. Generative AI tools were enabled for more than 110,000 employees by the Nov 2025 investor day, nearly 200,000 had AI tools on their desktops by the Mar 2026 annual report, which counted about 150,000 active users, the Jul 2026 earnings deck counted about 200,000 active users sending more than 400,000 prompts a day, and the Oct 2026 earnings release said more than 200,000 employees were using AI tools. Bank of America reported more than 600,000 AI training courses completed by Jul 2026. On the client side, Erica had 24.2 million active users in the third quarter of 2026, up from 20.4 million a year earlier.
By the numbers
Rollout, first pilot to today
Jun 2018
Bank of America launched Erica, its AI virtual financial assistant, in its consumer mobile app.
2020
Bank of America launched Erica for Employees, an internal virtual assistant for technology support, which expanded to human resources topics in 2023.
Aug 2023
Bank of America added Erica technology to CashPro Chat for business, commercial and corporate clients.
Apr 2025
Bank of America said more than 90% of employees used Erica for Employees and its developers saw efficiency gains of over 20% from a generative AI coding tool; Aditya Bhasin was Chief Technology and Information Officer.
Aug 2025
Erica passed 3 billion client interactions and nearly 50 million users since launch, and Hari Gopalkrishnan became Chief Technology and Information Officer.
Sep 2025
Bank of America launched AskGPS for Global Payments Solutions staff and named Dean Athanasia and Jim DeMare Co-Presidents, with the expansion of AI-based tools for clients in their remit.
Nov 2025
Bank of America's investor day reported generative AI tools enabled for more than 110,000 employees, more than 270 AI and machine learning models, and about 18,000 developers using GitHub Copilot.
Mar 2026
Bank of America's 2025 annual report said nearly 200,000 employees had AI tools on their desktops, about 150,000 were active users, and an AI catalyst program had launched a few months earlier.
Apr 2026
Brian Moynihan said all of Bank of America's roughly 200,000 employees have access to AI and about 90 AI installations were working.
May 2026
Brian Moynihan said 4,000 to 5,000 Bank of America employees were using Salesforce's Agentforce, growing by thousands a month as it spread to more businesses.
Jul 2026
Bank of America reported about 200,000 active AI users sending more than 400,000 prompts a day, more than 300 approved AI use cases and 114 live generative AI use cases, 34 of them fully deployed.
Sep 2026
Brian Moynihan said all of Bank of America's roughly 209,000 employees had access to AI, 130 to 150 use cases were implemented and about 50 more came on each quarter.
Oct 2026
Brian Moynihan said about 130 AI projects were implemented at Bank of America, each with a payback, and that technology development had become 10% more efficient.
Economics & controls
Spend and ROI
Technology budget
Bank of America spent $13 billion on technology in 2025, more than $4 billion of it on new technology initiatives, according to Brian Moynihan's Mar 2026 letter, and said in Jul 2026 that it spends $14 billion a year on technology with more than $4 billion for new initiatives including AI. Bank of America has spent more than $100 billion on technology over the past decade, more than $1.5 billion on data capabilities since 2021, and about $1 billion a year on cybersecurity (Nov 2025 investor day). Bank of America's own figures for annual technology spending vary between $12.7 billion and $14 billion across 2025 and 2026 documents.
What AI itself costs
Bank of America does not break out AI spending within its technology budget. Alastair Borthwick listed AI among the third quarter 2026 investments that raised expenses, alongside technology modernization, client acquisition and marketing. Brian Moynihan said in Sep 2026 that Bank of America expects to double its AI expense budget in 2027, and that the way it negotiated its licenses has so far kept token costs from becoming a problem. On Oct 14, 2026 Moynihan called AI a major part of the bank's growth investment and gave a figure of $250 million to $300 million in 2026 rising to $700 million in 2027 for projects in the pipeline, without saying whether it measured spending or benefits; CNBC reported the figures as savings.
Reported ROI
Brian Moynihan said on Sep 14, 2026 that Bank of America had 130 to 140 implemented AI projects costing $400 million and generating $800 million of benefit, which the bank was still in the process of capturing, and that each idea is costed with its own payback. Moynihan said on Oct 14, 2026 that about 130 projects are implemented, that their benefits reach two to three times their cost over time, and that none of them is a huge contributor on its own. Bank of America has not said whether the $800 million is annual or cumulative and did not repeat the figure on its Oct 2026 earnings call. Alastair Borthwick said in Jul 2026 that AI-enabled tools embedded in operations, risk, finance, technology and client-facing work have reduced manual work, in the same remarks in which he raised Bank of America's guidance for how far full-year revenue growth would outpace expense growth to 300 to 400 basis points.
Productivity markers
Bank of America says about 18,000 Global Technology software engineers gained more than 20% efficiency in code development from AI (Mar 2026), while Brian Moynihan said in Sep 2026 that the bank's 18,000 coders had saved 10% to 15% and put the gain across technology development spending at 10% in Oct 2026; Jim DeMare cited coding productivity gains of 15% to 20% for 19,000 to 20,000 developers (Sep 2026). Erica for Employees has cut IT service desk calls by more than 55% since 2020, EricaAssist cuts average call time by nearly one minute (Jul 2026), and Bank of America's investor day estimated that AI meeting preparation and summaries cut advisers' time on those tasks by about 90%.
Headcount
Bank of America employed about 213,000 people at Dec 31, 2025 and 2024, and 209,911 at Sep 30, 2026, down from 213,384 a year earlier. Alastair Borthwick said in Oct 2026 that headcount fell as the bank automated processes and simplified work, so it does not need to refill as many roles, and Brian Moynihan said managers fell about 5% while client-facing production staff grew. Moynihan said in Apr 2026 that Bank of America hires about 1,300 people a month to stay level and reduces headcount through attrition, and the Nov 2025 investor day showed consumer headcount falling from 101,000 in 2011 to 55,000 in 2025 on a slide about technology and AI.
Governance and risk
AI council and responsible AI framework
Bank of America's 2026 proxy statement says the company deployed generative AI tools with governance from an internal AI council, and Thong Nguyen's remit includes AI governance. Bank of America says its generative AI tools operate under a Responsible AI framework with governance, testing and oversight, and that employees keep responsibility for judgment, decisions and client interactions (Sep 2026).
Model risk review
Bank of America's Enterprise Model Risk Policy requires Model Risk Management to validate models independently before first use and to monitor and revalidate them, overseen by the Enterprise Model Risk Committee. Bank of America's 2025 Form 10-K says models face emerging risks from applications that rely on AI, and that the bank has internal processes and governance frameworks for AI-related risks.
Data boundary
Brian Moynihan said in Apr 2026 that Bank of America keeps customer data out of the models it uses, bringing models in without feeding them, and that protecting customer data has gated its adoption of some technologies, as it did with cloud computing.
Outside agents and agentic commerce
Bank of America and five other banks published joint principles for agentic commerce on Sep 22, 2026, covering transparency, safety, privacy and data, choice and interoperability. Brian Moynihan said on Oct 14, 2026 that Bank of America is working with AI agent providers and aggregators so that customers stay in control and a human approves before money leaves an account.
Board oversight
Bank of America's board expanded its strategy discussions on AI and brought in internal and outside experts on AI in 2025, according to the 2026 proxy statement. Bank of America added Bill Thomas, the former KPMG global chairman, to its board on Oct 21, 2026, and the bank's director biography says his experience strengthens the board's oversight of technology and AI enablement.
Disclosed risks and cyber
Bank of America's 2025 Form 10-K lists risks from third-party AI models with limited visibility into training data, inaccurate output, exposure of confidential data and autonomous AI agents that may act without authorization. Brian Moynihan told CNBC on Oct 14, 2026 that Bank of America added about 1,000 contractors in the second quarter of 2026, on top of about 3,000 staff in that area, to patch software flaws exposed by AI models including Anthropic's Mythos. Moynihan said third quarter 2026 remediation for that work and for anti-money-laundering orders cost a couple of hundred million dollars plus, and that the costs should run down.
Recent signals
Oct 21, 2026
Bank of America named former KPMG global chairman Bill Thomas and Dow executive chairman Jim Fitterling to its board, noting that Thomas oversaw a KPMG investment program focused in part on technology and AI-enabled capabilities.
Oct 14, 2026
Brian Moynihan said about 130 AI projects are implemented at Bank of America with benefits of two to three times their cost over time and that technology development is 10% more efficient, and Alastair Borthwick said headcount fell nearly 3,500 in a year as the bank automated processes.
Oct 14, 2026
Brian Moynihan told CNBC that Bank of America added about 1,000 contractors in the second quarter of 2026 to patch software flaws exposed by AI models including Anthropic's Mythos.
Oct 6, 2026
Evident's 2026 AI Index for Banks ranked Bank of America seventh of 50 banks, up three places, seventh on talent and eighth on innovation.
Sep 28, 2026
Bank of America launched the AskGPS Intelligence Hub, adding AI analysis of client, account and relationship data to the AskGPS assistant used by nearly 3,000 payments employees.
Sep 22, 2026
Bank of America, ASB Bank, Capital One, Commonwealth Bank of Australia, ING and NatWest published joint principles for trusted agentic commerce.
Sep 14, 2026
Brian Moynihan said at a Barclays conference that 130 to 140 implemented AI projects at Bank of America cost $400 million and generate $800 million of benefit, that about 50 use cases come on each quarter, and that the bank expects to double its AI expense budget in 2027.
Sep 2026
Bank of America's Commercial Bank launched a generative AI tool for prospecting and client onboarding, Alastair Borthwick said on the third quarter call.
Jul 21, 2026
Bank of America added generative AI to EricaAssist, used by more than 18,000 customer service representatives, and said it spends $14 billion a year on technology.
Jul 17, 2026
Bank of America named Kevin Milsom head of platforms AI transformation to drive AI adoption across its Global Markets group, according to a memo from Ashok Krishnan, head of platforms.
Jul 14, 2026
Bank of America added an AI slide to its earnings deck showing about 200,000 active users, more than 400,000 prompts a day, more than 300 approved use cases and more than 600,000 AI training courses completed.
May 27, 2026
Brian Moynihan said 4,000 to 5,000 Bank of America employees were using Salesforce's Agentforce, growing by thousands a month, and that 22 colleagues led by Jeff Busconi run the AI catalyst program.
Apr 15, 2026
Brian Moynihan said all of Bank of America's roughly 200,000 employees have access to AI and that the bank keeps customer data out of the models it uses.
Mar 26, 2026
Merrill and Bank of America Private Bank rolled out AI-Powered Meeting Journey at full scale for meeting preparation, notes and follow-up.
Mar 2026
Bank of America's 2025 annual report said about 150,000 employees actively use AI tools and that Brian Moynihan had launched an AI catalyst program with champions across the company.
Quick answers
- How does Bank of America deploy AI?
- Bank of America deploys AI through Erica, the virtual assistant it launched in 2018 and reuses for clients, advisers and staff, plus general AI tools that more than 200,000 employees were using in Oct 2026 and role-specific tools for advisers, bankers, call center staff and developers. Bank of America had more than 300 approved AI use cases in Oct 2026. CTO and CIO Hari Gopalkrishnan builds the technology, and strategy head Jeff Busconi runs the AI catalyst program that pushes usage and value.
- What does Bank of America use AI for?
- Bank of America had more than 300 approved AI and machine learning use cases in Oct 2026, with 114 live generative AI use cases and 34 fully deployed in Jul 2026, across consumer service, wealth management, banking, payments, markets, operations and software development. Bank of America has published a business return for two of them, AI fraud detection and AI coding assistance. For AI as a whole, Brian Moynihan said in Sep 2026 that 130 to 140 implemented projects cost $400 million and generate $800 million of benefit still being captured, and in Oct 2026 that the projects return two to three times their cost over time, with none of them a huge contributor.
- Who leads AI at Bank of America?
- AI at Bank of America is led by Hari Gopalkrishnan (Chief Technology and Information Officer); Jeff Busconi (Head of Corporate Strategy and Operational Excellence; AI catalyst); Thong Nguyen (Vice Chair and Head of Global Strategy & Enterprise Platforms). Ownership of AI on the org chart: CTO and CIO Hari Gopalkrishnan builds; Jeff Busconi drives usage as AI catalyst.
- What AI platform does Bank of America use?
- Erica, which Bank of America built in-house and launched in 2018, now also powers Erica for Employees, EricaAssist, ask MERRILL and CashPro Chat, alongside in-house generative AI tools such as AskGPS and Orchestra.
- Which AI models does Bank of America use?
- Bank of America gives staff multiple open-source and proprietary models in cloud environments it does not name, uses AI built into vendor tools such as Microsoft Copilot, GitHub Copilot and Salesforce Agentforce, and keeps customer data out of outside models (Apr 2026).
- How many BofA employees use AI?
- About 200,000 Bank of America employees were active users of AI tools in Jul 2026, sending more than 400,000 prompts a day, and 24.2 million clients actively used Erica in the third quarter of 2026.
- How far has AI spread at Bank of America?
- Bank of America sits at Scaling: all of its roughly 209,000 employees have access to AI tools, more than 200,000 were using them in Oct 2026, more than 300 AI use cases were approved by Oct 2026, and Jeff Busconi is accountable for AI usage and value as the company's AI catalyst. Bank of America falls short of Embedded on the financial test: its only firm-level return is Brian Moynihan's Sep 2026 conference estimate that 130 to 140 projects costing $400 million generate $800 million of benefit still being captured, and he said in Oct 2026 that none of the projects is a huge contributor. Usage is close to the Embedded bar, but Bank of America has not defined an active user or said how often employees use AI.
Open questions
- Which companies supply the open-source and proprietary models behind Bank of America's generative AI tools, and which cloud providers or data centers host them?
- Do the $250 million to $300 million for 2026 and $700 million for 2027 that Brian Moynihan cited in Oct 2026 measure AI savings or AI spending? CNBC reported them as savings, while Moynihan had said in Sep 2026 that the AI expense budget would double in 2027.
- Is the $800 million benefit Brian Moynihan cited for Bank of America's implemented AI projects in Sep 2026 annual or cumulative, and how much of it has the bank captured?
- How much of Bank of America's $14 billion technology budget goes to AI, including model and token costs?
- How many AI and machine learning specialists does Bank of America employ, and where do they sit?
- Who chairs Bank of America's internal AI council, and whom do Jeff Busconi and Hari Gopalkrishnan report to?
- How does Bank of America define an active AI user, and how many employees use AI tools every week?
- What business return do deployed tools with no published figures deliver, such as Erica for consumers, Erica for Employees, CashPro Chat, EricaAssist, AI-Powered Meeting Journey, AskGPS, Orchestra and document intelligence?
- How many of the more than 300 approved use cases are in production, and how do they relate to the 130 to 150 implemented projects Brian Moynihan cited?
- How much of the 3,500 fall in headcount from Sep 2025 to Sep 2026 came from AI, as opposed to other automation and process work?
- How much of Bank of America's third quarter 2026 remediation cost went to patching software flaws found by AI models such as Anthropic's Mythos, as opposed to anti-money-laundering work?
- Will Bank of America let customer-authorized outside AI agents act on accounts, and on what terms?
Sources
Bank of America
- Management team and directors
- Hari Gopalkrishnan, Chief Technology and Information Officer, biography
- Jeff Busconi, Head of Corporate Strategy and Operational Excellence, biography
- Thong Nguyen, Vice Chair, biography
- Dean Athanasia, Co-President, biography
- David Tyrie, President, Marketing, Digital and Specialized Consumer Client Solutions, biography
- Tom Scrivener, Chief Operations Executive, biography
- AI adoption by BofA's global workforce (Apr 8, 2025)
- Erica surpasses 3 billion client interactions (Aug 20, 2025)
- Senior leadership changes, Co-Presidents named (Sep 12, 2025)
- AskGPS generative AI assistant launched (Sep 30, 2025)
- CashPro Chat and transaction search reach record usage (Oct 27, 2025)
- Datos Impact Awards for AI and digital capabilities (Oct 28, 2025)
- CashPro Forecasting in 2025 (Dec 11, 2025)
- AI and digital innovations fuel 30 billion client interactions (Mar 10, 2026)
- Merrill and Private Bank launch AI-Powered Meeting Journey (Mar 26, 2026)
- EricaAssist enhanced with generative AI (Jul 21, 2026)
- Global banks collaborate on principles for trusted agentic commerce (Sep 22, 2026)
- AskGPS Intelligence Hub launched (Sep 28, 2026)
- Payments Insights launched for CashPro clients (Sep 28, 2026)
- Jim Fitterling and Bill Thomas named to the board (Oct 21, 2026)
Filings, letters and investor materials
- Q3 2026 earnings call transcript (Oct 14, 2026)
- Q3 2026 earnings press release (Oct 14, 2026)
- Q3 2026 earnings presentation (Oct 14, 2026)
- Q2 2026 earnings call transcript (Jul 14, 2026)
- Q2 2026 earnings presentation, slide 20 on AI (Jul 14, 2026)
- Q1 2026 earnings call transcript (Apr 15, 2026)
- Brian Moynihan at the Barclays Global Financial Services Conference, transcript (Sep 14, 2026)
- Jim DeMare at the BofA Securities Financials CEO Conference, transcript (Sep 23, 2026)
- Brian Moynihan at the Bernstein Strategic Decisions Conference, transcript (May 27, 2026)
- 2025 Annual Report and Brian Moynihan's letter (Mar 2026)
- 2026 Proxy Statement (Mar 2026)
- 2025 Form 10-K (filed Feb 2026)
- 2025 Investor Day, full presentation (Nov 5, 2025)
- 2025 Investor Day, payments, technology, operations and digital (Nov 5, 2025)
- 2025 Investor Day, welcome and firm overview (Nov 5, 2025)
News
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
Bank of America profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 1, 2026
