AI investments face tougher scrutiny over returns & risk

Technology and retail leaders are using AI Appreciation Day to warn that investment in artificial intelligence now requires clearer strategy, stronger guardrails and measurable business returns. Executives from Node4, Fluent Commerce and Syspro said AI projects are moving from experimentation into a more hard-headed operational phase.

Across manufacturing and retail, senior leaders say AI deployments now face tougher questions from boards and finance teams. They point to rising scrutiny over cost, data governance and dependence on specific model providers, along with pressure to show gains in efficiency and decision-making.

Manufacturers are expanding their use of AI across factory and back-office systems, with the focus shifting away from pilots and proofs of concept. Instead, organisations are embedding AI into production planning, inventory, quality control and maintenance as they navigate fragile supply chains and tight labour markets.

Syspro says this shift is most visible in industrial environments. Dan Abramson, Senior Vice President, Americas, Syspro, said manufacturers now treat AI as part of core plant operations rather than a separate innovation effort.

"This is the fifth AI Appreciation Day, and it feels remarkably different from the last. In just 12 months, AI has moved from being a technology the manufacturing industry was exploring to one it is now broadly embedding into critical day-to-day operations. Today is a chance to reflect not just on the extraordinary pace of AI advancement, but on how much manufacturers have adapted and achieved in putting it to work.

"Manufacturers continue to manage supply chain disruptions, labour shortages, tighter margins and changing customer expectations every day. AI is helping them cut through that complexity by making operational data easier to interpret, surfacing risks earlier and reducing the time teams spend on manual analysis.

"AI Appreciation Day also serves as a reminder that AI must serve people. The manufacturers seeing the strongest results are using AI embedded in the systems their people already rely on. Whether it supports production planning, inventory management or quality processes, AI is giving teams better information and more confidence to make faster, smarter decisions.

"Manufacturing has always depended on expertise, sound judgement and accountability. Every decision has implications for production, customer commitments, quality and safety. At its best, AI supports and strengthens those core principles, keeping teams at the centre of what matters most: quality, safety and customer commitment," Abramson said.

ROI Under Scrutiny

Retail and eCommerce operators are seeing similar trends. Early enthusiasm for generative AI and automation is giving way to closer examination of whether projects reduce costs or improve service.

Abdelkader Keddari, Director, EMEA South, Fluent Commerce, said many retailers are now focusing on targeted initiatives around stock, fulfilment and call centres as budgets tighten.

"After several years of rapid AI adoption, many businesses are entering a more cautious phase. The initial excitement led to widespread experimentation, but organisations are now under growing pressure to show that their AI investments are delivering measurable returns. As costs mount, AI needs to bring genuine operational value.

"Simply deploying AI is no longer enough. The organisations seeing the greatest return on investment from AI are those focused on solving specific operational challenges, such as streamlining customer service or improving inventory allocation, rather than implementing AI without a clear strategy.

"With AI investment likely to face even greater scrutiny, the focus should be on targeted projects with clear outcomes. Businesses also need to confront the reality that AI models will continue to evolve, providers will change, and geopolitical considerations may affect technology choices across markets.

"We have already seen moves in the US that could restrict access to certain models in Europe. In this environment, retailers need to avoid locking themselves into a single AI ecosystem and instead build flexible architectures that allow them to adopt different models as needs change.

"The competitive advantage will not come from backing one AI provider over another. It will come from having the operational foundations and data needed to take advantage of whichever innovations deliver the best result," Keddari said.

Foundations And Guardrails

Technology providers argue that many enterprises still move too quickly into AI pilots without first addressing data quality and governance. They warn this can lead to fragmented tools, regulatory exposure and limited operational impact.

Jay Fitzhenry, Virtual Chief Technology Officer for AI, Node4, said organisations need more rigorous planning before scaling automation and generative tools.

"AI adoption may be moving beyond the hype phase, but too many businesses are still rushing into projects without the right foundations in place. Clear planning is essential if they want to move beyond experimentation and achieve a genuine return on AI investment.

"We are already seeing some of the benefits AI has to offer. However, simply bolting it onto existing processes is unlikely to deliver more than incremental gains. Businesses need to be clear about which problems AI is best placed to solve and where it can drive meaningful change.

"Once the initial goals and desired outcomes have been identified, businesses must build robust guardrails and strong data governance into AI infrastructure from the outset. That means understanding exactly what data AI systems can access, how that data is used and where human oversight is required. Without this, businesses will not unlock the true value of AI.

"Even with the right foundations in place, people remain a critical part of the equation. AI can automate repetitive tasks and accelerate analysis, but it cannot replace human judgement. That is why human-in-the-loop approaches will remain essential, providing the oversight needed to validate outputs and step in when the technology produces an unexpected result, especially when decisions carry significant operational, ethical or reputational risk.

"Make no mistake, AI can deliver measurable impact, but only when implemented thoughtfully and responsibly. Otherwise, businesses will keep throwing money down the drain chasing the latest AI trend," Fitzhenry said.