From individual employees' grassroots attempts to a shared list of priorities

The group already had approved rules for working with AI and one corporate tool, but adoption was grassroots and uncoordinated, and licences were underused. Ideas came up in departments, but there was no process to collect them, assess them and route them to a pilot. The board needed to know which use cases would deliver the most value with the available data and at low risk.

The programme is ongoing. Its current outcome is a structured set of use cases for assessment, rather than verified results from deployed agents.

With the AI Champions network we ran six sessions on tools and use cases in FMCG and two workshops for collecting and assessing ideas. We run one-to-one consultations that produce idea cards. We agree materials and tools with the client's IT and security teams, and for all employees we prepared AI fundamentals training material compliant with the AI Act. The next stages are presenting the priorities to the board and workshops in five departments.

Company data goes only to the Copilot approved in the group, and in other tools participants practise on general data. Champions collect ideas in departments using shared material, and refined ideas go into the roadmap as cards. Scoring on five criteria (repeatability, business impact, data availability, pilot feasibility and risk) sorts ideas into quick wins, data-dependent projects, strategic initiatives and ideas that do not need AI at all. One-to-one consultations check feasibility, including GDPR constraints and integration with the group's systems.

Microsoft CopilotClaudePower BIn8nMake