AI training for membership organisations

Mostly for the work around the work: drafting and adapting written material, summarising long documents, cleaning and analysing data, preparing board and committee papers, and translating content. The organisations getting the most value have moved from individual experimentation to shared use cases per team, with simple guardrails. The biggest wins come from frequent, time-consuming tasks rather than from anything exotic.

Treat it as a head start, not a problem. Hidden, uneven AI use is the normal starting point in almost every organisation we meet. The leadership job is to make it visible and safe: agree which data can and cannot go into which tools, surface what is already working, and turn individual habits into team capability. Banning tools rarely works; it just pushes use further underground.

Start with a simple data classification: what is public, what is internal, and what is personal, confidential or otherwise sensitive. Match each category to approved tools and settings (enterprise agreements, EU data residency, no training on your data). In most organisations the majority of day-to-day AI use touches no personal data at all, so a clear traffic-light rule unlocks confident use of the safe majority while protecting the sensitive minority. Governance done well is a green light, not a brake.

The honest answer: the one your organisation can govern and afford, used well. All three can deliver real value, and they behave differently. What matters more than the choice is testing prompts on your real tasks, using organisational accounts rather than private ones, and building shared prompt libraries so good practice spreads. We train tool-agnostic and tailor to whatever your organisation runs.

Short enough that people actually read it. One page covering: which tools are approved, what data may go into them, where human review is required, and who to ask when unsure. The best policies are written as permissions rather than prohibitions, and they are taught, not just published. If your people can explain the policy in their own words, it is working.

Less than you think, more deeply than you expect. A single session should aim for one durable change, not six topics. Prioritise hands-on work on participants' real tasks, a simple rule for data safety, and one prompting habit that immediately improves output quality. Leaders additionally need a framework for judging AI use cases and a plan for leading adoption in their teams.