M&T Bank is not a household name outside upstate New York, but it is not small either. The Buffalo-based bank holds 212.9 billion dollars in assets, employs about 22,000 people, and runs more than 900 branches across 12 states and Washington, D.C. It has also spent the better part of a decade quietly rebuilding how it does technology, and that rebuild is now the reason it can move fast on AI.
The change started in 2018 when Michael Wisler took over as chief information officer after years at Capital One, a bank that went through its own tech overhaul a decade earlier. Wisler's main move was pulling technology work away from outside contractors and bringing it inside the bank. Today, more than 80 percent of M&T's tech and data staff work directly for the bank, up from a much lower share when he started.
This is not just an M&T story. Banking research firm Boston Consulting Group found that major banks now aim to have roughly 70 percent of their overall tech staff in house, with the remaining 30 percent or so coming from outside contractors. M&T is ahead of that curve, which is one reason it can roll out new tools faster than banks still leaning on outside vendors for basic technology work.
The payoff is visible in how M&T uses AI today. More than 15,000 employees now use AI copilot tools, which work like a smart writing assistant, to draft reports, summarize customer service calls, and help write software code. The bank is also testing AI that can watch for fraud and flag risk in customer accounts before problems show up.
None of that works without clean data, and M&T treats that as the real foundation. Before pushing AI further, the bank spent years training staff on data governance and mapping where its information actually lives. Wisler puts it simply: you cannot win with AI if you do not know where your data is or whether you can trust it. That is the part most companies skip when they rush to add AI tools, and it is usually why those tools disappoint.
There is a local angle too. M&T has poured more than 100 million dollars into tech centers in downtown Buffalo and backs local programs like TechBuffalo and the startup accelerator 43North, which puts 5 million dollars a year into the region's tech scene. The bank needed a steady supply of tech workers, so it helped build one, and the whole city benefits from that spillover.
The takeaway for any traditional business is not about banking software. It is about sequence. Insourcing the right skills and fixing your data comes before AI tools produce anything useful. Companies that skip straight to buying AI subscriptions without doing that groundwork will get faster versions of the same messy output they already have. M&T's advantage was built over years, not weeks, and that is exactly why it will be hard for slower-moving competitors to copy quickly.