By Musidora Jorgensen, UK & Ireland Country Leader, Freshworks
In July’s Mansion House speech, former Chancellor Rachel Reeves called AI the defining technology of our generation and used the occasion to unveil a wave of financial services measures built around it. From a new AI Economics Institute to backing the UK’s first agentic payments launch with Mastercard, the previous government positioned financial services as the sector best placed to lead on AI. Whether the new government continues that vision or not, the same distinction between adoption and execution must be made when it comes to AI software in the finance sector.
The most AI-bullish sector is also the most overwhelmed by it
Financial services firms are stewards of other people’s money, operating under some of the heaviest regulatory scrutiny of any industry. And yet they are also among the most enthusiastic adopters of AI anywhere. Recent research found 90% of financial services organisations are actively using AI, with nearly half planning to significantly increase investment over the next 12 to 24 months. That’s the highest rate of any sector in a global survey of more than 12,000 IT decision-makers.
It seems unusual then, that financial services are also the industry reporting the most severe AI-related complexity of any sector we surveyed. Organisations across all industries are losing an average of a quarter of their AI budget to complexity annually. But, in financial services, that number soars up to 40% as firms describe AI associated complexity as “very significant,” much higher than expected sectors healthcare and professional services.
Why financial services go at it alone
As eighty-six percent of firms claim that AI complexity has increased their team’s workload, part of the explanation is structural. Burdened by regulatory requirements, data sensitivity, and audit obligations, the financial services sector is most likely to build AI capability in-house rather than buy it. That instinct makes sense, but it’s an enormous contributing factor to complexity. In-house builds tend to create fragmented systems and governance models that are hard to unwind later.
I see this pattern constantly in conversations with lenders, particularly those serving customers with poor or limited credit histories, who are often least able to absorb operational friction. Vanquis’s recent move to Freshworks is a good example of the alternative path to rushing modernisation under regulatory scrutiny.
Pressure to deliver is outpacing the ability to deliver cleanly
Across financial services, IT leaders face intense, often unrealistic pressure to prove AI investments are paying off almost immediately, despite the technology’s inherent need for time to mature and deliver real value. Executives push for fast, visible wins while ignoring the practical realities of implementation. This mismatch has created a climate of genuine anxiety where eighty-four percent of respondents say their career progression feels at risk if AI doesn’t deliver within twelve to twenty-four months. That means that for most IT leaders, AI is now deeply personal and tied to their livelihoods.
Complexity is a choice, not a certainty
Trying to scale across disjointed in-house systems driven by personal pressures is bad for customers and the business bottom line. When Bangor Savings Bank consolidated its support function onto a single AI-powered platform, it achieved a 95% SLA resolution rate. Insurer Amerisure replaced siloed legacy systems with one platform across IT, legal, underwriting and HR, saving over 4,000 hours in 2025 and cutting onboarding resolution time by 97%.
The two goals that so often seem to pull in opposite directions, adopting cutting-edge AI capabilities and maintaining a simple, streamlined operating model, aren’t actually in tension with one another. If anything, the opposite is true. Real, sustainable AI adoption depends on having simplified, well-governed operations to build upon, while genuine operational simplicity increasingly depends on the smart, disciplined use of AI to achieve it.