M&T Bank Expands AI Copilots to 15,000 Employees After Tech Overhaul
M&T Bank AI adoption has expanded significantly, rolling out AI copilots to more than 15,000 employees in one of the most extensive enterprise AI deployments among US regional banks to date.
The bank now uses AI across internal operations, customer service, software development, and risk management, according to Fast Company. Specific applications include analyzing call-center conversations, drafting reports, generating code, identifying customer needs, and flagging portfolio risks. M&T is also exploring agentic AI applications specifically for cybersecurity and fraud detection going forward.
Why M&T Bank AI Access Was Restricted Before Expanding
American Banker reported in September 2025 that roughly 16,000 of M&T’s 22,000 employees were already using Microsoft Copilot for tasks like drafting emails, summarizing reports, and reviewing call-center transcripts. That widespread adoption didn’t happen overnight, though.
Before rolling out enterprise AI broadly, M&T initially blocked employee access to public large language models entirely. Chief data officer Andrew Foster explained that the restriction existed because employees could potentially input sensitive company information into public-facing AI services without adequate safeguards. The bank later evaluated enterprise-grade providers and selected Microsoft Copilot, starting with a pilot program involving roughly 800 employees before expanding access company-wide.
According to Foster, using generative AI to summarize call-center conversations saves approximately six minutes per call, a meaningful efficiency gain at the scale of a major regional bank. Software developers also rely on GitLab tools to generate code, though employees remain formally responsible for reviewing all AI-generated output before it’s used.
That human-review requirement isn’t just informal practice. It’s written directly into M&T’s 2026 Code of Business Conduct and Ethics, which mandates that employees use only approved AI tools and explicitly prohibits entering confidential, proprietary, customer, employee, or regulated information into any unapproved system.
Building the Technology Foundation First
M&T’s AI rollout didn’t happen in isolation. It follows a broader technology overhaul that began back in 2018, when more than half of the bank’s technology specialists were external contractors rather than in-house staff. Today, that ratio has flipped dramatically, with roughly 80% of the technology workforce now working in-house across more than 300 agile teams and roughly 2,000 technologists total.
The bank has replaced dozens of aging platforms during this transformation. Technology outages have dropped more than 80% since 2018, while the number of system upgrades completed annually has increased by 300% over the same period. Technology spending exceeded $1.2 billion in 2025, nearly triple what the bank spent in 2017, and annual technology releases jumped from roughly 15,000 in 2018 to 65,000 in 2025.
The Data Infrastructure Behind M&T Bank AI
Alongside the technology overhaul, M&T built out a parallel data-lineage program designed to track where information originates, how it’s used, and how it moves between internal systems. Foster, who joined the bank in 2023 to lead this effort, said the initiative wasn’t created specifically in response to generative AI, but rather as a foundational capability for understanding the bank’s broader data estate.
That groundwork has since become directly useful for AI deployment. M&T created an internal knowledge repository called Edison containing authoritative policy documents and institutional information, and uses data-lineage software from Solidatus and Monte Carlo to trace information as it flows through databases, applications, and business intelligence systems. According to Foster, this lineage work gives the bank clear visibility into the source, meaning, quality, and governance of individual data elements, which directly supports how Copilot and other tools access information through retrieval-augmented generation built on internal, governed data.
Three Paths to Scaling AI Across the Bank
M&T’s technology leadership described pursuing generative AI through three distinct pathways: general employee use, AI capabilities already embedded within existing applications, and proprietary systems built specifically around the bank’s own data and processes.
The bank operates more than 1,800 applications, many supplied by third-party vendors, and one strategic pathway involves identifying AI capabilities already built into those existing tools rather than building everything from scratch. The third pathway, proprietary development, has focused early efforts on repetitive operational work, software development, fraud prevention, and cyber defense specifically.
How Other Major Banks Compare
M&T isn’t alone in scaling generative AI across employee workflows. JPMorganChase launched its internal LLM Suite platform to more than 200,000 employees in 2024, and by 2025, more than 65,000 employees within its Corporate and Investment Bank division were actively using the platform, with over 90% of its engineers relying on AI coding assistants regularly.
JPMorganChase also reported that AI-based transaction screening allowed the bank to review more than twice its previous transaction volume while cutting manual operator checks in half.
Bank of America, meanwhile, has deployed a system called EricaAssist to more than 18,000 customer service employees. The tool summarizes why a customer is calling, retrieves relevant account information, and recommends next steps, while keeping the human employee ultimately responsible for the interaction itself. As of July 2026, Bank of America said EricaAssist delivers contextual guidance in under three seconds and has cut average call times by nearly a full minute, with plans to expand the system into additional service scenarios later this year.