As AI takes on more work, employees need new skills

As Microsoft expands what Copilot can do inside an organisation, the question of whether employees are equipped to use it becomes more pressing. Grant Thornton's AI Enablement team sheds a light on why literacy and adoption now matter more than licences.

For many employees, using Copilot still means asking it to summarise an email thread, draft a document or help prepare a presentation. These are useful applications, but Microsoft has much bigger ambitions for the technology. Whether organisations get their money's worth will depend in part on how well employees can put these new capabilities to use.

Microsoft’s latest announcement introduces new ways for employees to delegate work, build simple applications without coding expertise and use AI agents that can continue working after they have logged off. Some capabilities are still being tested or rolled out, but they point to a different way of working with AI.

Helping employees write better prompts is one thing. Preparing them to delegate tasks, supervise agents and judge when AI should act independently requires a broader set of skills. Consider the difference between asking Copilot to summarise a document and assigning an AI agent a series of tasks. In the first case, an employee reviews the response before deciding what to do next. In the second, they may need to decide what information the agent can access, what actions it can take and when a person must step in.

The challenge for organisations is keeping employees' skills up to date as the tools change. AI training can no longer be treated as something completed at rollout.

Where to start

A useful assessment of AI readiness begins with the work employees perform and the decisions they make, rather than the number of Copilot licences deployed or how frequently the tool is opened. 

Three questions tend to surface first:

1. Do our employees understand which AI capabilities are relevant to their work, beyond summarising and drafting?
2. Do we know which workflows would genuinely benefit, and have we asked the people who run them?
3. Who inside the organisation is equipped to support colleagues as the capability keeps developing?

The answers can help identify where employees need further training, which processes merit closer examination and where internal support may be lacking. A finance team using Copilot to prepare management commentary may have different requirements from an IT team developing automated workflows. Employees handling sensitive information may also face restrictions that limit which tools or functions they can use. 

Training should account for these differences. Introductory sessions can establish basic knowledge and useful working habits, while applied training can help employees evaluate specific processes and identify where AI could reduce manual effort or improve the quality of work.

More advanced programmes may need to address agent supervision, application development, automation and governance. These require an understanding of how AI systems behave across multiple steps, what can go wrong and where human intervention remains necessary.

Building AI literacy into everyday work

An effective AI literacy and fluency programme should combine foundational skills for all staff with applied training tailored to their workflows. IT teams and internal champions may need deeper technical training to help colleagues adopt new capabilities and resolve problems as they arise.

Grant Thornton's AI Literacy & Fluency Programme, for example, is designed around these different levels of need. Programmes are tailored to an organisation's starting point, with practical assignments and assessments intended to connect learning to everyday work. Preparation and follow-up activities help participants apply what they have learned beyond the classroom.

Baseline assessments and structured feedback can help organisations measure changes in participants' confidence and identify areas requiring further support. However, these measures alone cannot establish whether training has improved how employees work. Organisations also need to understand whether participants are applying their skills and whether those applications are producing useful results.

As Copilot takes on more complex work, organisations will need to decide which capabilities are relevant to their employees, where additional training is required and how that learning translates into better ways of working.