Banks and financial services firms accelerating software delivery with the help of generative AI tools may be heading into what Andrew Power calls the “AI speed trap.”
The head of UK & Ireland at software testing firm Tricentis said that, while AI is supercharging code creation and release velocity, quality assurance is failing to keep pace, with potentially costly consequences for banks and insurers under regulatory and operational pressure.
“In the rush to capitalise on Generative AI, software development and delivery has shifted into overdrive. Teams are moving faster, delivering more code, and automating everything from testing to deployment,” London-based Power wrote in a recent blog.
“In many ways, it’s a golden age for software productivity. But beneath the surface, a growing problem threatens to undo those gains: software quality isn’t keeping up,” he explained.
Rising risks
Power warned that “the more we trust AI to ship code autonomously without rigorous due diligence, the wider the quality gap becomes. And the consequences are already visible. Outages, security breaches, mounting technical debt, and, in some cases, millions in annual losses as the result of business disruption.”
Citing recent data, he added: “Recent research shows that two-thirds of global organizations are significantly at risk of a software outage within the next year, and almost half believe poor software quality costs them $1 million or more annually. There is an emerging tension in AI-driven software development: speed vs. stability.”
Modern DevOps and CI/CD pipelines already prioritise velocity, but according to Power, generative AI has “turbocharged this further, creating more code than ever before.”
However, “AI doesn’t ensure quality; it ensures output. We all know that AI can get it wrong. Without proper guardrails, AI-powered development becomes like a high-speed factory churning out code without accountability. So why are so many teams pushing code live without fully testing it? Because the pressure to deliver quickly outweighs the mandate for due diligence.”
He pointed to findings from the 2025 Quality Transformation Report, noting: “Nearly two-thirds of organizations admit to releasing untested code to meet deadlines. It’s a staggering statistic, and a stark warning.”
Redefining quality
Power argued that traditional QA metrics are being distorted: “Traditional metrics like test coverage, defect rates, or system stability used to define quality. Today, speed is starting to stand in for quality, but it’s a dangerous substitution. Shipping faster doesn’t mean shipping better.”
“If quality becomes synonymous with velocity, teams risk ignoring deeper indicators, including resilience, maintainability, and customer experience. And when those things fail, the fallout can be major: lost revenue, compliance failures, or service outages that damage trust,” he said.
“Software quality must be redefined for the AI-first world. It’s not just about finding bugs, it’s about ensuring long-term performance, user satisfaction, and business continuity. In this landscape, quality is less about the absence of errors and more about the presence of confidence.”
“Why are so many teams pushing code live without fully testing it?”
– Andrew Power
Despite accelerated pipelines, many teams remain hesitant.
“Over 8 in 10 of EMEA IT teams, as well as 73% in the U.S., say they delay launches because they aren’t confident in their test coverage. The disconnect between external pressure to move fast and internal uncertainty about product stability is a symptom of broken feedback loops and incomplete visibility,” Power said.
“This breakdown isn’t just a technical issue; it’s a cultural one. To fix it, organizations need stronger alignment around goals, clearer quality metrics, and smarter automation that doesn’t just accelerate work but elevates it.”
Power emphasised that while AI adoption is growing, oversight must remain human-led: “Autonomous AI agents making release decisions may boost productivity, but without transparency, explainability, and traceability, they can also introduce risk at scale. Responsible AI use in development means embedding governance into automation. It means having a way to audit what AI did, and why.”
“This starts with AI-literate teams. Developers and testers need to understand the logic behind AI-generated outputs, not just blindly accept them. Ethical awareness, systems thinking, and contextual judgment must be part of every team’s toolkit if AI is going to serve as a true partner in quality.”
Closing the quality gap
For Power, closing the gap requires strategic investment in QA. “If software engineers want sustainable gains from AI, leaders need to clearly define what quality means for their teams, what level of risk is acceptable within their business and build that into testing strategies from day one.”
“The quality gap won’t close with more speed, but with smarter systems. This means investing in autonomous software testing and quality intelligence not as an afterthought, but as a strategic function,” he said.
By embedding AI-driven insights and real-time automation into the software lifecycle, financial services firms can “deliver at speed without compromising reliability.” B
ut, Power concluded, fundamentals remain critical: “Clear requirements, continuous feedback, and cross-functional accountability. These aren’t outdated concepts, they’re the foundation for any resilient development practice. In short: if AI is the engine, quality must be the brakes and the steering.”
NEW EVENT
Why not become a QA Financial subscriber?
It’s entirely FREE
* Receive our weekly newsletter every Wednesday * Get priority invitations to our Forum events *
REGULATION & COMPLIANCE
Looking for more news on regulations and compliance requirements driving developments in software quality engineering at financial firms? Visit our dedicated Regulation & Compliance page here.
READ MORE
- Inside Rabobank: Engineering resilience by design
- Can AI agents finally automate data testing?
- Continuous testing drives DORA compliance
- Why software testing may face a major rethink
- Buy or build? AI rewrites software testing for banks
WATCH NOW
