DVT, an expert-led, AI-driven software and data engineering services company, recently brought together enterprise leaders and practitioners to bridge the gap between AI marketing hype and practical software delivery. The Insights session, titled: "Hype vs reality: Smarter software development with AI", focused on why structured oversight and expert leadership are now more critical than the tools themselves.
Opening the event, Ronnie Cloete, DVT chief customer officer, set a tone of pragmatism, urging enterprises to move past "informal experimentation". Cloete emphasised that for AI to deliver sustainable value in software delivery, it must be underpinned by the same discipline and accountability as traditional engineering.
Saurabh Agrawal, CTO at DVT, presented a framework for evaluating AI in software engineering and challenged common industry narratives, including claims that AI replaces engineers, delivers 10x productivity or eliminates QA functions. He noted that productivity gains in well-run environments typically fall within a 20%-40% range.
Agrawal also examined the economic implications of AI adoption, explaining that usage costs, model selection and monitoring introduce new oversight requirements. He stressed that AI adds a new cost layer to software delivery, requiring financial visibility and engineering-level governance rather than informal experimentation.
“The goal is not to be pessimistic about AI, but to be economically precise,” says Agrawal. “The 10x productivity claim is a vendor narrative. AI doesn’t eliminate engineering teams. It shifts effort away from repetitive tasks. Unchecked AI code in production is a liability, not a saving.”
Adding a layer of practical application, Richard Basson, DVT engineering lead, shared internal benchmarks. DVT is currently utilising AI to draft specifications and accelerate prototyping, but Basson was quick to emphasise that human validation remains the non-negotiable "guardrail" to prevent over-reliance on AI-generated output.
Enterprise perspectives from Standard Bank reinforced the importance of structure and accountability. Gladys Hlokomayo, head of technology research at Standard Bank, addressed myths around the so-called “death of engineering”, emphasising that AI augments skilled teams rather than replacing them. She stressed that AI investment must be tied to measurable business outcomes, with human accountability maintained across the life cycle.
Building on this theme, in his session: “Who’s Driving?” Richard Hudson, lead software engineer at Standard Bank, explored how developers are evolving into AI orchestrators, remaining in control while co-ordinating agents within structured delivery frameworks rather than relying on ad hoc tooling.
A consistent theme across the session was that AI in software engineering delivers value when embedded within clear governance models, validation checkpoints and strong engineering leadership.
Cloete concluded that AI alone is not the answer. “AI is powerful, but without experienced engineering leadership it creates risk as quickly as it creates speed,” says Cloete. “Expert oversight, governance and accountability are what turn experimentation into production-grade delivery.”
The session concluded with peer discussion and networking, with further DVT Insights events planned for 2026.
Share
DVT
Founded in 1999, DVT is an expert‑led, AI‑driven software and data engineering services company headquartered in South Africa. With in-country presence across Africa, Europe, the UAE and Australia we provide our customers with global scale and local presence to deliver real business value from technology.
DVT’s services include custom software engineering, AI solutions, digital transformation, data and analytics, automation, cloud services and UX/UI design.
DVT is part of the Dynamic Technologies group. For more information, visit www.dvtsoftware.com.
Editorial contacts