Many organisations begin their AI programme by switching on Copilot and giving everyone access, expecting productivity to follow. Access is the easy part. Turning it into measurable value is where most programmes stall, and it is the risk we see organisations underestimate most.

Value risk gets less attention than security risk

Security and privacy risks are well understood and widely managed, and they should be. Value risk gets less attention. A tool available to everyone and used by few for anything that changes an outcome produces activity without results, and the investment keeps running regardless.

Training works when the pathway is already clear

In a recent programme with a client, we ran an organisation-wide adoption and capability uplift. Little of the time went on what AI is or how it developed. The training was anchored to the organisation’s strategic direction, its use cases and the operational changes it was making, and people learned prompts and tools inside the processes they work in every day.

The organisation had also settled the pathway before training began: policies, approved technologies, and data and security rules. When people returned to their jobs, they could use what they had learned straight away and practise it in the moment.

The opposite pattern is common. An off-site, a large training session, then a return to day jobs where the AI cannot be used the way it was taught.

Measure the whole process

AI applied to one step usually lifts that step. Whether the organisation sees the gain depends on the steps around it. If the work before and after the improved step stays the same, the time saved is absorbed there, and the result across the whole process is a fraction of the local gain. MIT NANDA’s 2025 GenAI Divide report found that only about 5% of integrated generative AI pilots extracted measurable business value.

Value shows up when the whole function or process is reimagined with AI in mind: who does what, which handovers remain, and what the business measures. The measure that matters is the outcome of the process from start to finish, owned by someone accountable for it.

AI becomes part of the transformation agenda

Boards and executives are still asking what the organisation’s AI plan is. That is healthy, and without deliberate design it produces a lot of poorly invested dollars. Over the next year or two we expect AI to become less of a topic and more of an ingredient. The initiative is a transformation of how work happens, enabled by AI, automation and data, and it has to be sustainable for the business and for the people in it.

Questions to ask before the next rollout

• Which use cases is this rollout anchored to, and which process does each one change?
• Are policies, approved tools, and data and security rules settled before training starts?
• Can people use what they learn in their day job straight away?
• Which whole process will we measure, and who owns the result?
• What changes in the surrounding steps so the saved time shows up?

Not sure how to get value from your Copilot rollout

The Omnia Collective works with financial services and enterprise organisations across Australia as a certified Microsoft Solutions Partner, helping teams turn AI access into measurable value through use-case-led capability uplift and change across the whole process. See our AI and automation services or get in touch.

This post is adapted from William’s appearance on Episode 21 of Nap Stack, Monica’s podcast on AI, data, and building a business. Listen on Spotify or Apple Podcasts.