Leadership
AI for Departments & Leaders
A six-stage roadmap for introducing AI in a department or education organisation — honestly, safely and at a pace your team can sustain. No promises of measured learning gains or time savings: those depend on your context, your tools and your people.
It gives you structure: audit, agree, evaluate, train, prepare learners, review. It does not replace your organisation’s governance, procurement, safeguarding or data-protection processes — it works through them. Formal tool approval stays with whoever holds that authority in your organisation.
The six stages
Stage 1 · Audit where you are
Survey existing AI use (formal and informal), staff confidence, current policies and learner access. Output: a one-page snapshot of where the department actually is.
Stage 2 · Agree priorities and boundaries
Agree two or three priorities — no more. Set boundaries, name the decision-makers, and define learner rules. Output: a department AI position, in writing.
Stage 3 · Evaluate tools properly
For each candidate tool check: age requirements, data handling and whether a data protection impact assessment is needed, accessibility, cost, impact, exit plan, and staff testing. Output: an approved tools register — formal approval only, from the people authorised to give it.
Stage 4 · Build staff capability
Plan CPD (the free CPD directory helps), identify champions, embed inclusive practice, set up peer support. Output: a termly CPD plan.
Stage 5 · Prepare learners
Learner AI literacy, clear assessment expectations, safeguarding routes, and access alternatives for learners without devices or connectivity. Output: a learner guide.
Stage 6 · Review and adjust
Review confidence, time, learner feedback, incidents, changing tool terms and what to improve. Output: a review record — and the next cycle begins.
Running the roadmap
Give every stage a named owner, a due date, a status and a defined output. The action plan spreadsheet below does exactly that. Expect a full cycle to take a term or more — faster usually means stage 3 was skipped, and stage 3 is where the data-protection and safeguarding risks live.
Example: FE department, six weeks in
A curriculum department starts with an anonymous staff survey (stage 1). It finds 60% already use general AI tools informally, and nobody knows the organisation’s position. The department’s first priority becomes turning informal use into approved use — not adding new tools. Stage 2 produces a one-page position: which tool, for what, what’s never typed into it, and who approves changes.
Common mistakes
- Skipping the audit — you can’t govern what you can’t see.
- Approving tools by conversation instead of process.
- Buying a tool before defining the problem it solves.
- Training staff but not informing learners.
- No exit plan when a tool changes its terms, price or age limits.
- Treating a one-off review as “done”.
Troubleshooting
| Problem | Try |
|---|---|
| Staff resistance | Start from workload problems staff name themselves; make use optional and supported. |
| Policy doesn’t mention AI | Flag to your data protection officer and leadership; interim position in the meantime. |
| Tool terms changed | Re-run stage 3 evaluation; check the exit plan. |
| Nothing changes after CPD | Pick one workflow, one owner, one review date. |
Five questions before approving any new AI tool
- Data: what happens to information entered into it, and does it need a data protection impact assessment?
- Accessibility: does it work with assistive technology, and is it accessible to every learner who’ll use it?
- Cost: what is free, what is paid, and what happens to price after the trial?
- Learner safety: what are the age limits, moderation and reporting routes?
- Exit: if the organisation stops using the tool, how is its content and data retrieved?
The leadership pack
Page last reviewed: 7 November 2026 · Next review: by May 2027 · No claims of measured learning gains or time savings are made — outcomes depend on context. See Research, Guidance & Evidence for dated official sources.