Bank of America plans to double its expense budget for artificial intelligence next year as the lender seeks measurable returns on its technology investments. Co-President Jim DeMare said during a BofA Securities conference appearance that investors are pressuring companies to show faster results from AI spending. The Charlotte, North Carolina-based bank is prioritizing productivity gains over experimental projects, focusing on areas where the impact is clear.
Measurable gains in software development
DeMare noted that the most significant AI returns in financial services so far are found within technology units. The bank’s 20,000 software developers are using coding agents to optimize their workflow, boosting coding productivity by about 15% to 20%. This gain is consistent with what the company sees across most industries, and DeMare described the improvement as “clearly identifiable.”
BofA is also using its AI-powered virtual assistant, Erica, to handle internal self-service requests. This effort has led to a considerable reduction in help desk inquiries, with the bank stating Erica handles the work of about 11,000 people. The technology helps employees access information without needing to submit a formal request to support staff.
Expanding the budget and headcount
CEO Brian Moynihan said last week at a Barclays conference appearance that the bank expects to double its AI expense budget for the upcoming year. He cited about 140 implemented uses, noting the initiatives cost $400 million while generating a benefit of $800 million. The lender currently spends about $4 billion annually on new tech initiatives, with AI being a growing part of that total.
Despite the increased investment in AI, the bank’s headcount has actually decreased from about 213,000 employees at the beginning of the year to 209,000. Moynihan said the attrition rate is about 8.5%, and the bank is managing hiring carefully rather than laying off workers. He emphasized that employees are held accountable for AI-generated information they use, ensuring that the technology supports human oversight rather than replacing it entirely.
Employee-driven innovation and guardrails
DeMare mentioned that about 95% of the company has access to AI tools, which has increased familiarity and comfort. He said the focus is on general productivity and how AI improves workflows. Looking ahead, Gopalkrishnan is evaluating the level of autonomy the bank is comfortable giving AI agents. He noted that as models become more efficient, the bank will expand autonomy but will wait for appropriate guardrails before doing so.
