Banks and other financial services providers are often seen as flagbearers for AI adoption. With digital-first experiences now firmly established as the baseline, organisations throughout this sector are exploring how AI can accelerate decision-making, sharpen insight and improve customer outcomes.
While experimentation in these areas is widespread, operational impact is not. So, as financial institutions move beyond the AI proof-of-concept stage, the true challenge becomes whether or not AI can perform reliably and securely at scale.
A new way of defining performance
This shift is already encouraging business leaders to take a performance-first stance on AI. Take, for instance, a metric like the efficiency ratio, which has traditionally offered a backward-looking assessment of cost versus income – evaluating if a resource input was worth the output it produced.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Now, leading banks are beginning to use that same metric as a forward-looking indicator of AI readiness – a sign of its effectiveness for reducing operational friction and delivering smarter outcomes without added complexity or risk. With that, the question changes from the retrospective “did this investment pay off?” to the anticipatory “can our digital environment support AI at scale without breaking under the strain?” For many, the answer remains “not yet”.
High confidence, low readiness
The ambition is clearly there – more so than in most other industries – but studies reveal that growing tension at the heart of AI adoption in the banking industry is down to investment and confidence not being matched by operational maturity.
In fact, according to the 2025 Riverbed Global Survey, only 40% of organisations in the financial sector consider themselves ready to operationalise AI. Just 12% of AI initiatives are fully deployed enterprise-wide, while 62% remain stuck in pilot or development phases.
These results expose a readiness gap that must be bridged before AI projects can viably scale – particularly now the FCA is actively reviewing AI’s role within retail banking. Early pilots can demonstrate potential, but they’re simply not designed to deliver sustained value. Success will be determined by an organisation’s ability to transform those isolated initiatives into AI that works continuously across the enterprise.
US Tariffs are shifting - will you react or anticipate?
Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.
By GlobalDataThe barriers holding AI back
There are two main obstacles consistently preventing AI from progressing into everyday banking operations:
- Operational complexity – The majority of financial institutions operate across hybrid and multi-cloud environments layered with legacy systems, third-party platforms and distributed networks. Deploying AI within this fragmented landscape is inherently difficult.
- Limited visibility – Without a unified view of the whole digital estate, AI-driven services might behave unpredictably. Issues take longer to diagnose, resolution times increase and operational risk rises – particularly when customer-facing or revenue-critical services are hampered.
If unaddressed, a live digital environment like this – one that consists of inflexible legacy systems with ineffective monitoring tools – leaves little room for successful innovation or compliance and, ultimately, prevents AI from becoming the “transformation silver bullet many are claiming it to be”.
Taking control of the complexity
To move AI pilots smoothly into production, banks must create a frictionless operating environment as possible. One way to take proactive action is through unified observability, which provides a real-time, holistic overview of applications, infrastructure, networks and user experiences – eliminating blind spots and giving decision-makers clarity over performance and risk.
Using that advanced visibility as a foundation, AIOps solutions can then correlate signals, reduce alert noise and accelerate incident resolution through intelligent automation – helping IT teams to address any bottlenecks or performance-restricting issues before they impact the end-user.
Other ways to scale AI operationality include standardising on OpenTelemetry, making infrastructural changes that accelerate data movement, and investing in data quality and governance to ensure AI remains compliant. By embracing solutions that do all of the above, decision-makers gain a shared, real-time understanding of how systems behave – and, crucially, how AI performs in practice.
Illustrating the effectiveness of AI
As deployment alone isn’t a meaningful indicator of value, financial sector leaders must also reconsider how success is measured as AI matures.
For instance, with the help of an observability platform, banks can stop customers from detecting critical payment failures before the system does – and as a result, they should also see a measurable reduction in Mean Time to Resolution for transaction-related incidents, alongside fewer escalated helpdesk tickets.
Clearly, tracking all relevant outcomes before and after deployment provides tangible evidence for AI’s long-term impact. When positive results are sustained, organisations can feel reassured that their investments are genuinely reducing friction and supporting their people – rather than adding yet another layer of complexity.
From optimism to optimisation
The debate about AI’s place within the financial services sector has already been settled. The emphasis is now on preparing for innovation to graduate to execution. In order to operationalise AI at scale, organisations should embrace solutions that simplify infrastructure complexity, support real-time decision-making and provide superior digital experiences to all end-users.
The institutions that don’t take a proactive stance will be left with little to show for the pilots that were initially considered causes for optimism. On the other hand, those that confidently move on from experimentation will be far better positioned to protect service continuity and meet regulatory standards as they grow. They’ll be able to use their measurably improved performance to unlock even more competitive advantages.
Fernando Castanheira, Chief Information Officer, at Riverbed Technology
