From shadow AI on personal accounts to a ranked plan, a compliance grid and a first AI policy.
To protect confidentiality, we may alter specific details while preserving the accuracy of our core contribution.
The client
A Belgian environmental non-profit with about 100 employees, running around 12,000 workshops a year with 3,700 partner schools.
The challenge
The organisation faced more administration, falling subsidies and a growing need for fundraising. Many staff already used AI, often on personal accounts, with no common framework or quality standard, and opinions on AI split the team. The leadership team needed a shared understanding of AI and a plan it could stand behind.
What we did
This was a pro bono mission under our Data for Good programme, run like a paid engagement.
A 12-question questionnaire before the training, to see where each participant stood
Eight individual discovery interviews with members of the leadership team
A plenary training session: AI fundamentals, the difference between uncertainty and manageable risk, prompting, live demonstrations, AI agents, limits and risks
Individual 90-minute coaching sessions
A cross-analysis of the interviews: 16 AI opportunities scored on impact and effort, a three-wave action plan over about six months, a GDPR and AI Act compliance grid and a benchmark of tools
A check of the 16 opportunities against interview quotes and an expert challenge
Support for the organisation’s first AI use policy
Results
A shared vocabulary on AI across the leadership team, including staff who were sceptical
16 opportunities ranked, with a plan in three waves: quick wins, data and process clean-up, then strategic bets
A compliance grid that tells teams which uses are fine, which need care under GDPR and the AI Act, and which are off-limits
A first AI policy drafted by the organisation, with our support
What we learned: AI adoption here was a change question more than a tool question. Clean data in the CRM and document storage came before any ambitious use case.
Technologies
Generative AI assistants (training and demonstrations), impact and effort scoring, GDPR and AI Act compliance grid.
To safeguard confidentiality, we may modify certain details within our case studies.