Anand Naidu is a pivotal figure in modern industrial software development, bridging the gap between raw code and factory floor efficiency. With a profound understanding of how software quality dictates the success of autonomous systems, he brings a unique perspective on the intersection of AI and manufacturing during this period of rapid digital transformation. His expertise spans the full stack, allowing him to see both the microscopic bugs in the code and the macroscopic impact they have on global production chains.
This discussion explores the rapid acceleration of AI adoption within manufacturing workflows, highlighting the inherent tension between deployment speed and software governance. We delve into the mounting financial risks associated with inadequate testing, the “tool sprawl” phenomenon currently hindering progress, and why quality engineering has transitioned from a backend IT concern to a fundamental pillar of industrial infrastructure. The conversation also addresses the critical skills gap and the physical consequences of software failure in an era of digital twins and connected robotics.
The latest data reveals a striking contradiction: while 85 percent of manufacturers trust AI agents to handle software deployment autonomously, only 31 percent feel truly prepared to govern these systems. How does this gap between confidence and capability impact the stability of a production environment?
This discrepancy creates a fragile foundation where the enthusiasm for innovation outpaces the safety nets required to catch errors. When you have nearly nine out of ten leaders willing to let AI pull the trigger on a software release without human oversight, you’re looking at a massive shift in operational trust. However, without the governance frameworks to manage that autonomy, we see a “black box” effect where decisions are made but not understood or controlled. In a factory setting, this isn’t just about a website crashing; it’s about a robotic arm or a digital twin receiving faulty instructions that could halt a production line. Bridging that 54 percent gap between trust and readiness is the most urgent task for leadership right now to avoid catastrophic oversights.
With 57 percent of manufacturing organizations admitting to deploying code changes without full testing, what are the primary pressures driving these risky shortcuts in the development cycle?
The pressure is palpable and usually comes from two directions: the boardroom and the sheer volume of output. About 32 percent of manufacturers explicitly point to leadership pressure to move faster, which creates a culture where “done” is prioritized over “perfect.” At the same time, 29 percent of teams are simply overwhelmed by the sheer amount of software that needs testing, a problem that AI actually makes worse by generating code faster than humans or traditional systems can verify it. It’s a frantic race where the finish line keeps moving further away, and the result is a “deploy now, fix later” mentality. This approach is dangerous because once that untested code is live in a connected machinery environment, the cost of a rollback is significantly higher than the cost of doing it right the first time.
Many organizations are struggling with “tool sprawl,” with 37 percent citing an overabundance of separate AI and automation tools as a barrier. How does this fragmentation complicate the quest for continuous software quality?
When you have a patchwork of disconnected tools, you lose the “single source of truth” that is vital for high-stakes manufacturing. Instead of a streamlined pipeline, you end up with data silos and compatibility headaches that actually slow down the very processes they were meant to accelerate. This fragmentation makes it nearly impossible to scale quality engineering because each tool requires its own specific expertise and maintenance. We also see that 35 percent of organizations are hitting a wall because of gaps in AI and machine learning expertise, and managing twenty different tools only thins that limited talent pool even further. To achieve true continuous quality, manufacturers need to move away from this “tool for every problem” approach and toward integrated platforms that provide a holistic view of the software lifecycle.
As software becomes more embedded in physical assets like industrial robotics and predictive maintenance systems, why must we stop viewing quality engineering as just an IT issue?
In the modern factory, software is just as much a part of the infrastructure as the steel and the electricity. When 34 percent of manufacturers identify security breaches or compliance failures as their top concern, they are recognizing that a software bug is now a physical vulnerability. If the software driving a predictive maintenance system fails, you don’t just lose data; you lose a multi-million dollar piece of equipment that wasn’t serviced when it should have been. We are seeing software quality move from the server room to the factory floor because, in an autonomous world, the code is the operation. If the software doesn’t behave as intended, the physical process it controls becomes unpredictable and potentially hazardous.
The financial stakes are clearly rising, with 63 percent of manufacturers estimating losses of over £372,000 annually due to poor software quality. What are the long-term consequences for those who fail to address these mounting costs and technical debt?
The financial bleed is significant, and for one in five companies, those annual losses can climb as high as £3.72 million, which is enough to derail any R&D budget. Beyond the immediate cash drain, you have 30 percent of organizations struggling with the weight of technical debt and maintenance costs that compound over time. This creates a “quality tax” on every new feature or innovation they try to implement, eventually leading to the 25 percent of companies that report slower release cycles due to constant rework. Perhaps most damaging is the erosion of trust with partners and customers; once you lose the reputation for reliability in an industrial supply chain, it can take years to recover. Those who don’t invest in quality engineering now are essentially mortgaging their future ability to compete.
What is your forecast for the evolution of autonomous software testing in manufacturing?
I expect we will see a mandatory shift toward “governance-first” AI implementation, where the ability to audit and control an AI agent becomes more valuable than the agent’s speed alone. Within the next few years, the role of the traditional software tester will transform into a Quality Engineer who manages fleets of AI agents, focusing on the strategic orchestration of testing rather than manual execution. We will see the rise of “closed-loop” systems where the digital twin not only simulates production but also continuously tests the software governing it in a real-time feedback loop. Ultimately, the manufacturers who thrive won’t be the ones who deployed AI the fastest, but the ones who built the most resilient frameworks to ensure their AI never operates in a vacuum of accountability.
