The traditional technology divide between an organisation’s business leaders and its engineers has been flipped in the artificial intelligence age, with end users and data consumers calling the shots in proposed pilots and making tougher demands of their software developers and data managers.
The role reversal is such that AI has become a business model issue rather than a technology one for financial institutions, according to Harsukhdeep Dahia, Senior Vice President and Head of Financial Services at quality engineering firm QualityAI.
And as AI deployment grows and the new development dynamic puts greater regulatory and reputational scrutiny on business leaders, the demand for defensibility and accountability in AI-led products and decisions has made data quality and the implementation of governance guardrails even more critical, Dahia told Data Management Insight.The London-based executive argued that the advent of AI and agentic workflows has revealed the vast potential of how technology can be used within financial institutions and other industries. With data democratisation meaning non-technical minds are now able to harness AI to build automated workflows with simple voice or written commands, innovation for specific use cases is now increasingly in the hands of data consumers.
“For the first time in my 30-year career, business are actually on par with engineers – if not, slightly ahead – of where technology needs to be going in order to serve their business needs, which is a really interesting juxtaposition to be in,” Dahia said.
“Historically, technology teams often led the exploration of what was possible. AI has changed that dynamic because business users can now participate much more directly in defining and testing what creates value.”
Blurred Lines
Formerly Qualitest, QualityAI provides quality engineering and data assurance services to clients in regulated industries. It works to enable them to scale their technological transformation across data platforms, cloud and AI. Dahia describes the company’s work as identifying where technology can meet business needs, helping them test the performance and quality of solutions, and building governance guardrails – or “trust wrappers” – to ensure clients’ AI initiatives are robust, defensible, explainable and compliant.
The blurring of lines between business and engineering teams is happening across the industry. Last month, for instance, Finbourne Technology said it had created a team of engineers with business skills that would be embedded within its clients’ teams to help them not only implement its tech but also guide them on how it can enhance their operations.Drawn from the engineering and sales teams, the “business change consultants” are charged with building AI agents as well as identifying where in their business agents could fit.
Dahia argued that where technology programmes become disconnected from business ownership, they risk solving the wrong problem, irrespective of how strong the underlying engineering is.”.
With business leaders in charge, governance, explainability, compliance and risk management would be “baked in” from the start. Projects were more likely to come to fruition as a result.
“The business has cottoned onto it because there’s an expense involved, there’s a value realisation involved. You’re no longer really experimenting to see how fast things can go or what productivity gains you can get because in order for it to be accepted and in order for the developer’s role to be justified, they need to make sure as an engineer that it’s actually been adopted out there,” Dahia said.
“If they don’t involve the business… they are literally creating a rod for their own back.”
Feedback Loop
There is a constructive feedback loop in the new landscape too, with the risk of failure now a deterrent to pursuing a faulty idea. In the past, Dahia said, projects might be run on only to justify the investment.
“Are you really going to put your name against it, your career against it, if things start to drift?” he said. “That’s why trust and explainability is so, so key. Regulators are still catching up a bit as well.”
The changed roles have also come about because the business side of organisations has become more savvy about costs, especially when it comes to AI usage. As AI moves from experimentation into production, the economics become much more visible and the business has to demonstrate that the value created justifies the cost of running and governing these systems, said Dahia.
This leads to a more mature and disciplined approach, where AI initiatives are increasingly judged by their safety, trust, explainability, and ability to meet regulatory standards, rather than just by the technical innovation of the model itself.
“It’s now more about looking at projects in a broader lens and making sure that there is value baked in upfront and there is trust and explainability is baked in upfront rather than later,” he added.
“It’s pay a little now or pay a lot more later because, guess what, if you take a poorly governed model into production, you risk regulatory intervention and remediation costs. . And if it’s not the regulator, it’s going to be a loss of user trust.”
Subscribe to our newsletter
