Artificial Intelligence is no longer a topic reserved for tech startups or Silicon Valley giants.
The American banking sector is proving it knows how to play this game too, and M&T Bank is one of the most interesting examples of that shift. 🏦
With over 150 years of history and roughly 22,000 employees spread across the United States, this regional bank isn’t exactly the kind of institution you’d expect to lead a technological transformation of this scale.
But that’s exactly what happened.
M&T Bank put AI copilots in the hands of more than 15,000 employees, integrating the tool into daily workflows across different departments, from customer service to software development, internal operations, and risk management.
And the most interesting part here isn’t just the scale of the project.
It’s the way everything was built with patience, strategy, and above all, responsibility when it comes to data usage.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Before releasing any AI tool to employees, the bank spent years restructuring its technology foundation, training teams, and establishing clear governance rules.
The result of that journey is what we’re going to dig into here. 👇
A solid foundation before any AI move
What stands out about M&T Bank’s trajectory isn’t the end product, but everything that came before it. While many companies rushed to adopt artificial intelligence tools without much planning, the bank chose to build the foundation before putting up any structure on top of it. That technology overhaul started back in 2018, when more than half of the bank’s tech specialists were contractors. Today, the picture has flipped: roughly 80% of the technology workforce is in-house, comprising about 2,000 professionals organized into more than 300 agile teams. During that program alone, the bank hired over a thousand specialists.
The numbers that come with this transformation are impressive. System outages and disruptions dropped by more than 80% since 2018, while the number of system updates completed annually grew by 300%. Technology investments also surged, exceeding $1.2 billion in 2025, nearly triple the amount recorded in 2017. To put the pace in perspective, annual technology releases jumped from around 15,000 in 2018 to a staggering 65,000 in 2025. This wasn’t an overnight flip of a switch — it was a gradual, well-thought-out process.
A big part of this progress has a name behind it. Wisler joined M&T Bank as chief information officer in 2018 and, by 2025, moved into the role of senior executive vice president of technology and operations, overseeing both the technical side and the bank’s operational functions. Alongside him, chief data officer Andrew Foster, who came on board in 2023, played a decisive role in building a reliable and traceable data foundation — something that proved essential when it came time to scale AI safely.
Data governance: the invisible foundation
A critical piece of this process was the investment in data governance. The financial sector deals with extremely sensitive information, and releasing an AI tool without clear guidelines could create serious compliance and privacy risks. It’s no coincidence that before large-scale adoption, the bank actually blocked employee access to public language models. As Andrew Foster explained, the concern was real: employees could unintentionally feed confidential company information into services open to the public. 🔒
That’s when the bank began evaluating enterprise vendors and chose Microsoft Copilot, starting with a pilot of about 800 employees before expanding access across the entire organization. A September 2025 report indicated that 16,000 of the approximately 22,000 M&T employees were using Copilot for tasks like drafting emails, creating reports, and summarizing call center conversations. And the gains are very tangible: using generative AI to summarize customer interactions saves roughly six minutes per call. That might sound small, but multiply it by thousands of daily calls and the impact is massive.
Foster built a data lineage program to track where information comes from, how it’s used, and how it flows between systems. Interestingly, he made a point of clarifying that this work didn’t start as a response to generative AI, but rather as a core capability for understanding the bank’s entire data estate. For this, M&T uses specialized software like Solidatus and Monte Carlo, which help trace the path of information through databases, applications, and business intelligence systems.
The bank also created an internal repository called Edison, which houses official documents and information about internal policies, along with a Data Academy focused on governance and data skills, with around 2,000 employees participating in the program. This entire structure enables the use of techniques like retrieval-augmented generation (RAG) with governed internal data, ensuring more accurate and secure responses.
How AI became part of the bank’s daily routine
With the infrastructure in place and teams trained up, AI started being woven into work routines across different areas. Wisler explained that M&T is pursuing generative AI along three main paths: general employee usage, AI features already embedded in existing applications, and proprietary systems built around the bank’s own data and processes.
In customer service, the tool began assisting with conversation analysis, response suggestions, and organizing interaction histories, which cut down average resolution times and freed up agents to focus on more complex situations. That kind of gain might seem small on its own, but when multiplied across thousands of daily interactions, the impact on business processes is pretty significant.
In software development, AI also proved its value in a clear way. Engineering teams began using GitLab tools for code generation, with one important detail: employees remain responsible for reviewing all AI-produced work. That requirement for human review isn’t just a best practice — it’s formally documented in the bank’s 2026 Code of Business Conduct and Ethics.
The policy is quite clear on this point. It requires employees to use only approved AI tools and prohibits entering confidential, proprietary, customer, employee, or regulated information into unauthorized systems. On top of that, it makes explicit that employees remain accountable for the accuracy and appropriateness of any work done with AI assistance. It’s technology serving people, not the other way around. 🚀
Applications that go beyond the basics
If early use cases revolved around writing, summaries, customer service, and software development, today M&T Bank is already exploring far more sophisticated applications. Among the newer initiatives are identifying customer needs and flagging portfolio risks, tasks that require a more analytical and contextual reading of data.
The bank runs more than 1,800 applications, many of them provided by third parties, and one of its strategic approaches is identifying useful AI features already built into those tools. Meanwhile, the proprietary development front focuses on repetitive operational work, software development, fraud prevention, and cyber defense. M&T is even eyeing agentic AI applications for cybersecurity and fraud detection, showing that this journey is far from over.
A movement that reflects the entire banking sector
The M&T Bank experience isn’t an isolated case. Other major American banks are also expanding their use of generative AI across workflows. JPMorganChase, for example, launched its internal LLM Suite platform to more than 200,000 employees in 2024. By 2025, over 65,000 employees in the corporate and investment banking division were actively using the platform, while more than 90% of engineers had access to AI-powered coding assistants. The bank also reported that AI-driven transaction screening allowed them to review more than double the previous volume, cutting manual checks in half.
Bank of America is following a similar path with its EricaAssist system, available to more than 18,000 customer service employees. The tool summarizes the reason for the call, pulls up relevant information, and recommends next steps, always keeping the employee in control of the interaction. According to the bank, as of July 2026, EricaAssist delivers contextual guidance in under three seconds and has reduced average call times by nearly a minute.
The most valuable takeaway from this story is strategic patience. In a world that celebrates speed above all else, M&T Bank chose to do things differently. It spent years building the right foundation, invested heavily in people before investing in tools, and only then accelerated large-scale adoption. That pace might look conservative at first glance, but in practice, it’s what ensured the bank’s artificial intelligence project became something sustainable, scalable, and aligned with the institution’s values — rather than just another pilot that fizzled out along the way. 💡
