TLDR
- Middle-market companies are adopting AI faster than expected as they look to improve efficiency and drive new revenue growth, according to a Key Commercial Bank survey.
- Many firms are moving beyond back-office automation and beginning to use AI for top-line activities such as sales, marketing, pricing and customer engagement initiatives.
- The bank itself is taking a cautious AI approach with strict guardrails, keeping humans involved in sensitive banking workflows while using AI to improve customer service and internal operations.
The pandemic years were tough for middle-market companies: Supply chains were disrupted, business resilience was tested and a big jump in inflation all conspired to usher in an existential crisis.
And yet, as these economic shocks hit them all at once, they found a way to not only survive but thrive: Revenue increased on average by double-digit percentages from 2021 to 2025, according to a recently released report from Key Commercial Bank, KeyBank’s commercial banking unit serving this market.
“They were really dynamic that way,” Ken Gavrity, president of KeyBank’s commercial banking unit, said in an interview with The AI Innovator. “They found a way to absorb the costs, and a lot of it was through investments in technology and operations.”
According to KeyBank’s survey, 77% had a positive outlook for their company’s performance over the next 12 months – nearly a historic high – even if only half have a positive view of the U.S. economy.
Why? “While economic uncertainty persists, company-level optimism has jumped significantly, driven by breakthrough gains in operational efficiency, a better labor market, and the potential of AI to unlock business value,” according to the survey of 750 financial decision makers.
So it may not come as a great surprise that middle market companies – defined by KeyBank as those with $10 million to $1 billion in annual revenue − are embracing AI faster than many large corporations, not just to cut costs but increasingly to drive revenue growth.
“They’re a bit more ready for this technology shock than I think people expected them to be,” Gavrity said. “Sixty-nine percent plan to continue to invest in the business and … top of the list is AI.”
“This technology is more pervasive than anything I’ve seen in my 30-plus years” in the industry, he added. “Whether it’s industrial companies, business services, accounting, law firms, consultancies – all of them are experimenting with it. … That is surprising for us to see in this cohort.”
Flexibility to outmaneuver larger rivals
Without the large coffers of their bigger rivals, middle-market companies have banked on adaptability – with AI’s help − to outmaneuver competition.
“They’re trying to figure out, how do I use this to my advantage?” Gavrity asked. While public companies move slowly as they set up governance models, middle-market firms are building proofs of concept and tapping third-party software for CRM, supply chain, content generation and other use cases.
“They’re going back to all those vendors to say, ‘how are you starting to embed artificial intelligence into your software? What’s relevant to me? … How do I deploy this?’” he said. “Show me how to use this.”
To be sure, middle market companies aren’t yet seeing “material” profit gains from AI but “early returns are positive enough that they’re saying, ‘I have deployment schedules I’m thinking about in 2026 and 2027 already,’” Gavrity said.
The survey showed that 51% are actively implementing AI and automation, which is a new top catalyst for growth. Three-quarters of companies plan to use AI to automate employee tasks, up 19 percentage points from a year ago, while 68% expect the main benefit to be productivity improvement. Also, 71% are using AI for data analysis.
The challenges they face are ensuring effective human-AI collaboration (51%), managing job security concerns (48%) and reskilling employees (45%). Cybersecurity is also a rising area of concern: 52% plan to increase spending moderately while 17% expect a “significant” hike.
For 2026, middle-market firms plan to implement AI faster, deploy capital offensively and treat cybersecurity as competitive exposure, not just IT risk.
AI moving the top line
The nature of AI experimentation is also evolving rapidly. Early deployments focused heavily on back-office productivity improvements, but companies are now beginning to test AI for revenue generation, including sales targeting, lead generation and personalized marketing campaigns.
“Where this started as a back-office, mid-office opportunity, you’re starting to see … that this can actually drive the front end,” Gavrity said. Revenue generation can come from better pricing strategies, customized offers for customers, digital nudges to encourage clients to sign up, among others.
Some manufacturers are using AI to optimize predictive maintenance for industrial equipment, while professional services firms are using generative AI tools to accelerate drafting and research work, Gavrity said.
KeyBank is also using AI, but selectively and with guardrails. Executives described a deliberately cautious but expansive AI strategy built around what they called a “two-pronged approach.”
One effort focuses on giving employees access to generative AI productivity tools such as Microsoft Copilot under strict internal controls. The other centers on carefully selected operational domains where AI can help reduce friction in customer service and internal workflows.
The bank has already rolled out Copilot companywide with extensive guardrails, said Kim Snipes, executive vice president and head of commercial digital, AI and client experience at Key Commercial Bank, in the interview. Employees are prohibited from entering personally identifiable information into AI systems, and usage is closely monitored.
“Our teams cannot just use whatever tool they choose,” she said. “We monitor that very closely.”
The bank internally describes its AI maturity framework as “guide, assist, do.” In the earliest stage, AI systems guide employees by surfacing insights and recommendations. Later phases may involve limited execution capabilities once workflows are well understood and sufficiently controlled.
Executives emphasized that KeyBank is not allowing AI systems to directly control sensitive banking functions such as money movement or account processing. Instead, early deployments focus on areas such as customer servicing, where AI systems can aggregate information from emails, phone calls and digital interactions to help employees understand customer issues more quickly.
One example involves using AI to consolidate fragmented interactions from commercial banking clients. “You would say, ‘You better know me. I call you all the time,’” Snipes said. “That’s where this type of capability brings it all together for us.”
Gavrity said the bank is intentionally keeping humans involved in critical workflows. “We’re years away from being able to fully remove a human from a process like that,” he said. “You’re not decoupling the human execution with the artificial intelligence insight.”
Be First to Comment