Proving AI deployment value needs a more strategic approach

You wouldn’t judge the value professional chefs bring to Michelin-starred restaurants by how quickly they chop onions. So why is there such a focus on time saving at work through AI usage?

AI tools, from copilots to agentic systems, have the potential to reinvent entire business workflows, impacting business models and changing how enterprises deliver value. Time savings, on the other hand, are easy to measure through employee surveys and analytics collected from heavily monitored activities such as customer care.

So taking this easier approach and just trying to do the same activities in less time is not the way to look at this. Longer-term strategic thinking is required, and businesses that examine what they do from the ground up, and consider how AI could allow them to reconfigure these activities to create value for customers, will be the winners in these turbulent times.

Cutting headcount is not a strategy

No company ever shrank to greatness. Sure, laying off staff in downturns may sometimes be necessary, and deploying AI as a way to do this may bring a short-term bump to the bottom line. But it’s missing the real opportunities AI presents and could actually do longer term damage to a business. Gartner predicts half of companies cutting customer service staff due to AI will rehire by 2027. They reason AI is just not mature enough to replace the subtleties and empathy that humans bring to customer interactions. The danger, Gartner argues, is that while immediate cost savings may be attractive, there’s a danger of unintended consequences further down the line.

This sentiment is echoed by Ronnie Sheth, CEO of SENEN, a strategic consultancy focused on data and AI. She warns against taking a technology-first approach such as deploying copilots and self-service chatbots without thinking about customer needs and preferences. It’s counterproductive, as a growing number of companies are finding, to handle more customer queries if customer satisfaction levels go down. “To truly realize value, AI initiatives should map directly to strategic business objectives by a defined percentage, not just internal soft metrics such as time saved,” Sheth says.

Focus on quality not quantity

In a world starting to drown in AI-generated slop, the quality of outputs and customer interactions will become ever-more important. Copilots and other AI tools can play a key role in driving quality up, but choosing the right metrics to measure this is key. Founder of technology advisory firm Deep Analysis, Alan Pelz-Sharpe sees the focus on deploying copilots primarily to raise employee outputs as a mistake. “Somebody might write more code, but is it good code and does it actually deliver measurable business value,” he asks. “So if your measurement is tasks completed per hour or headcount, you can argue that you did more, but did you do better, did you deliver enough value to justify the cost and change.”

On top of this, he warns of not monitoring for possible negative consequences further down the line. This is particularly true with coding assistants capable of producing code at speed but which, without skilled human oversight, can cause problems weeks or months later as systems become exposed to the harsh realities of business environments.

A 2025 survey by Harness of 500 software engineers found that while just over 80% felt AI was useful in their jobs, they were also struggling with a blast radius effect where bad code was going into production faster and with a greater impact. This included debugging with 67% of developers spending more time cleaning up code generated by AI, and 68% spending more time dealing with security vulnerabilities from such code.

Not business as usual

Despite justified talk of AI bubbles fuelled by some ridiculous investments and valuations of companies unlikely to ever live up to their promises, a revolution is underway. Seeing AI as a bolt-on that can help businesses perform a bit more efficiently is missing a key point. How data is gathered, processed, shared, and used is being transformed in terms of speed and quantity but, more importantly, AI is changing how decisions are made on the basis of that data.

As agentic solutions evolve, many roles currently performed by humans will either radically change or disappear altogether. When word processing systems emerged in the 1970s and 1980s, a key measure of their value was how much quicker secretaries and admin assistants could produce written documents. However, the real impact of word processors and the PCs that replaced them has been the transfer of much of this work to executive staff, who now manage their own admin tasks.

The impact of AI will be far greater. Enterprises need to rethink how useful their current business processes and, by extension, their business models will be when AI is able to automate or even make current practices redundant. Simply performing current activities quicker is not the way to think about this.