Microsoft Copilot: A Year of Learning

Mixed feelings from the frontline

According to recent research conducted by Freeform Dynamics in collaboration with the CIO WaterCooler, while around half of organisations have either deployed or are actively exploring Microsoft Copilot, impressions from early adopters have been decidedly mixed. Only 16% report being genuinely impressed with the technology, while 36% express outright disappointment. The remainder sit somewhere in the middle – seeing potential but remaining cautious about both value and enterprise-readiness.

“A bit hit and miss” was how one CIO summed it up, while another observed that it’s “Good, but needs a plan and strategy to get the most from it.” Perhaps most telling was the comment that “The license cost far outweighs the benefit,” something we’re hearing quite a lot at the moment.

Copilot as a microcosm

What’s particularly interesting about the Copilot experience is how it reflects the broader state of play with generative AI in the enterprise. The research suggests that most CIOs acknowledge the technology’s transformative potential, but few are convinced it’s ready for prime time. Indeed, when asked to characterise the current crop of generative AI solutions, just 9% considered them truly mainstream-ready, with the majority (55%) viewing them as suitable only for early adoption.

The unavoidable reality

But whether you’re convinced or not, generative AI is becoming increasingly difficult to avoid. Microsoft’s strategy of embedding Copilot capabilities across its product suite means that if you’re running a Microsoft shop, you’re going to have to deal with it one way or another. The functionality will keep appearing on users’ desktops through every possible route.

So what does this mean for IT leaders?

Learning to tame the beast

The research suggests that the most successful organisations are taking a structured approach rather than letting things evolve organically. This starts with getting the fundamentals right – particularly around security, compliance and governance. As one study participant put it: “Security is a major issue, especially given the limited experience in this area.”

Beyond this, there’s a strong emphasis on identifying concrete use cases that align with genuine business needs and have the potential to deliver real value. The “build it and they will come” approach, or what we might call the “chuck it over the wall and see what happens” strategy, is generally seen as risky and potentially counterproductive.

Looking ahead to 2025

Will 2025 be the year when generative AI truly goes mainstream in the enterprise? The jury’s still out on that one. What’s clear from the research, however, is that most CIOs are planning for a measured evolution rather than a revolution. This applies whether we’re talking about Copilot, other embedded AI capabilities, or more specialist generative AI deployments.

The bottom line

If there’s one key message coming through from UK CIOs, it’s the importance of maintaining control. By all means experiment and learn – that’s essential – but do it in a purposeful manner. The last thing you want is to end up footing the bill for functionality that creates more problems than it solves while delivering questionable value.

In practical terms, this means putting appropriate guardrails in place, then moving forward deliberately with clearly defined objectives. Microsoft may be pushing hard with Copilot, but that doesn’t mean you have to dance to its tune. Take time to understand the technology, its implications, and how it can genuinely help your business. The same principle applies to generative AI more broadly.

If you want to learn more about the research mentioned in this article, the full report entitled “Generative AI Checkpoint” can be downloaded from here.