By Klaus Kurz, Senior Director at New Relic
Financial services organisations are under intense pressure to deliver digital services faster while maintaining resilience, security and regulatory compliance. Research shows that 75 percent of UK-based financial services firms are already using Artificial Intelligence (AI), underscoring how deeply the technology is embedded in the sector. AI is also accelerating software delivery in financial services, with finance leaders ranking code generation as the highest-impact AI use case by a significant margin per Gartner.
New software releases are now more frequent, and traditional testing and governance processes are increasingly being outstripped. This is shifting risk into live environments and placing observability – the ability to understand system behaviour through real-time data from across applications, infrastructure, and user interactions – front and centre.
At the same time, tolerance for failure has all but disappeared, especially as the costs of downtime rise. Recent data shows that high-impact IT outages now cost financial services firms an average of $1.8 million (£1.3 million) per hour, turning system reliability into a direct financial risk rather than a purely technical concern. In a sector where digital platforms underpin payments, trading and customer service, even brief disruption can have immediate and visible consequences.
When risk moves into live financial systems
Research shows AI-generated code produces 1.7x more defects across key software quality categories including logic, maintainability, security, and performance. Additionally, the Bank of England notes that the complexity of some AI models, coupled with their ability to change dynamically, poses new challenges around the predictability, explainability and transparency of model outputs.
Long-established safeguards such as layered governance and human oversight were designed for a slower pace of change, where risk could largely be assessed before release. For financial services, where regulatory compliance and risk management are critical, these issues can create substantial operational and regulatory risks.
And when failures occur, they rarely remain isolated. Outages are often prolonged, with incidents lasting hours rather than minutes, allowing issues to escalate rapidly across interconnected systems. Nearly 29 percent of financial services organisations report high-business-impact outages at least weekly. These are not rare anomalies but recurring operational challenges.
The impact extends beyond disruption alone. Engineering teams now spend an average of 31 percent of their time responding to outages and incidents, diverting capacity away from innovation. In a sector defined by competition, customer expectations and regulatory scrutiny, this combination of frequent disruption and lost productivity compounds the cost of failure and increases operational risk.
A widening gap between change and control
Despite rising exposure, many organisations still lack the end-to-end visibility needed to understand how changes affect performance and customer experience once systems are live.
In financial services, customer trust and seamless digital interactions are essential. Recognising this, finservs are accelerating plans for stronger experience monitoring – 89 percent plan to deploy browser monitoring, 80 percent plan to deploy mobile monitoring and 77 percent plan to deploy synthetic monitoring over the next one to three years.
Yet AI monitoring deployment in financial services stands at just 50 percent, below the cross-industry average of 54 percent. While small, this gap is notable given that financial services operates in one of the most tightly regulated and risk‑sensitive environments, where observability and AI monitoring should be a top priority. Clearly, there is a growing mismatch between the pace of change and organisations’ ability to observe system behaviour once changes reach production.
Observability to enhance business outcomes
Against this backdrop, the ROI of observability needs to be reframed. With the right visibility, teams can detect early warning signs before customers are affected, whether through abnormal payment behaviour or performance degradation in digital channels. This allows organisations to intervene earlier and reduce the likelihood that routine changes become high-impact failures.
Observability delivers measurable business value. 53 percent of organisations say it helps mitigate service disruptions and business risk, while 42 percent of financial services organisations report a return on investment of two times or more.Whilst observability doesn’t improve code quality itself, it can directly increase how fast teams detect and resolve issues.
Why APM matters as AI-generated code scales
Application Performance Monitoring (APM), a core pillar of observability, is becoming increasingly important as the volume and speed of code increases. APM provides dashboards and alerts to troubleshoot application performance in production. These insights target known system failures – typically the four SRE golden signals of latency, traffic, errors, and saturation – and alert engineers to pre-defined issues with troubleshooting guidance.
Beyond helping teams instantly allocate code errors, APM surfaces areas where code is generally correct but runs inefficiently on its infrastructure. This inefficient code can drive up cloud costs unnecessarily, making performance monitoring essential for managing spend.
Seeing more to move faster
The use of AI-generated code will continue to grow across financial services. The challenge is whether risk controls evolve to match that reality. Embedding observability into delivery and operations allows innovation to scale with confidence, rather than accepting uncertainty as the cost of progress. In a sector where outages are highly visible and increasingly expensive, clear insight into live systems is essential to moving faster without increasing risk.