How AI is transforming IT in financial services

Financial firms were early to AI. Now, agents are changing IT’s job, and midsize firms feel it most.

Two employees looking at a screen with financial content
Dan Tynan

Dan TynanThe Works Contributor

Nov 16, 2026 • 5 MIN READ

The financial services industry has always lived on the cutting edge of technology. Decades before AI chatbots commanded the world’s attention, banks and trading companies were using machine learning to process loan documents, assess credit risk, and flag suspicious transactions.

AI agents are a different proposition. According to Evident Insights, nearly 1 in 3 new AI use cases in Q1 2026 involved agents, double the number in Q4 2025. Firms are using them to automate loan processing, fraud detection, wealth management, customer screening, and much more. That expands the number of end points and use cases that IT has to manage, and that sprawl is growing exponentially. In a single year, the number of financial services employees who have access to AI tools has doubled to more than 60%, according to Deloitte.

The largest institutions can staff dedicated AI platform and governance teams. Midsize banks, credit unions, insurers, and asset managers face the same regulatory scrutiny only with more limited resources. Here’s how that’s changing the tasks, technology, and teams in IT at financial services organizations.

Read also: The bottom line for trust in financial services? IT.

Tasks: Watching over the bots

Many IT teams are shifting from doing the work to overseeing the agents that do it, says Tim Burke, president and CEO of Quest Technology Management. That includes curating AI-generated knowledge, monitoring models for bias and drift, and maintaining logs for regulators.

Platforms that bring automation and audit trails together make that oversight workable. When Live Oak Bank adopted Freshservice to handle ITSM tickets, it found the same AI-powered solution could automate processes for nine other business units within the $15 billion bank. AI-based root-cause analysis helped engineers solve problems faster, while auditable logs recorded every action taken, meeting the bank’s compliance needs. Insurer Amerisure replaced siloed legacy systems across IT, legal, underwriting, and HR with a single AI-powered platform, saving more than 4,000 hours in 2025 and cutting employee onboarding resolution time by 97%.

40% of financial services respondents call AI-related complexity "very significant," 11 points above the overall average.

Technology: Managing agentic workflows

For financial services IT, the first question about agents isn’t which tool to pick. It’s whether to build or buy. In a Freshworks survey of more than 12,000 IT decision-makers, 43% of financial services respondents said their primary approach to AI is building rather than buying. That’s the highest of any industry, compared with 34% overall, 30% in manufacturing, and 27% in the public sector.

The reasons are specific to the industry. Because firms must keep tight control over their data, they’re keeping information assets in house and treating AI models as replaceable, says Ran Aroussi, co-founder of VarOps, which builds and deploys AI systems inside mid-market companies.

“Finance is more sensitive than most sectors to AI models being retired or changed,” he says. “A growing share of AI work in regulated firms is moving onto infrastructure they control, because the data it needs isn't allowed to move.”

Building also puts new work on IT, starting with identity and permissions for agents. “AI tools now call into company systems on people's behalf, so something has to decide, per request, who the human is, which AI client is asking, and what each is allowed to do,” he says. “That logic should live on the company's side, not inside each AI vendor's product. Otherwise, a firm with five AI tools ends up with five separate definitions of who can see what.”

The cost shows up in the data. Forty percent of financial services respondents in the Freshworks survey described AI-related complexity as “very significant,” 11 points above the overall average. Nearly one-third report that AI outputs are generating substantial rework, versus 29% overall. Only 41% describe their AI implementations as fairly well-orchestrated across systems and teams.

Read more: The most AI-bullish industry is also the most overwhelmed by it

Teams: Smaller teams, broader skills

Apart from AI roles, which have grown 24% since 2023, according to the Evident AI Index, IT teams at finance firms are not getting any larger. But the responsibilities they’ve been handed continue to increase, along with the breadth of skills IT pros are being asked to provide. In Freshworks’ survey, 86% of financial services IT leaders say AI-related complexity has increased their team’s workload.

Robert Joslin, managing partner at Avasant, sees IT organizations shifting from large monoliths to smaller multidisciplinary teams that combine domain expertise with engineering, architecture, cybersecurity, and AI skills. Rather than build everything internally, they lean on third-party providers for specialized capabilities. 

“This is especially relevant to midsize financial institutions, where fragmented legacy applications and limited specialist resources can restrict AI adoption,” he says.

That leaves the people who stay with a broader job. “Our financial clients used to hire pure infrastructure people,” says Timur Rakhmatullin, senior software engineer at Softline Solutions. “Now they need engineers who also understand access controls, data governance, and compliance.”

Governance: IT plays a crucial role 

For financial services IT, governance is the hard part of agents, and the part that’s harder to hand off. When an agent acts across several systems of record, each with its own governance model, someone has to be able to reconstruct what happened.

“There are very strict regulatory requirements about how financial data gets managed,” says Tim Crawford, founder of AVOA. “But a lot of the regulations that are in place aren’t designed for the dynamic nature of AI, and the tooling to truly manage governance at scale doesn’t really exist yet. The smartest companies are using AI for very specific purposes, so if something goes sideways, they can contain the damage.”

IT teams can play a crucial role in governance by limiting the types of regulated data AI models can see, says Aroussi. Instead of giving models access to everything at once, tech teams can enforce existing access controls at the moment of retrieval, so an AI agent only ever receives the information the person asking is allowed to see.

“That puts IT in the middle of AI governance in a way it wasn't before,” he says. “Identity, entitlements, and audit were always IT's job. AI just made them the main control, not a background one.”

Read more in this series on how AI is changing IT across industries: How AI is changing healthcare IT