02Education · Policy and change
Many schools, one way to use AI.
A multi-academy trust wanted its teachers to gain time from AI without putting pupils’ data or the integrity of assessment at risk.
- Client
- A multi-academy trust in England
- Scale
- Many schools, primary and secondary
- Duration
- Seven months, including a one‑term pilot
- Team
- Two from ATP, with a safeguarding specialist from our network
A representative example of our work. Client details are withheld.
The problem
What it was costing.
Teachers were already using AI, each in their own way. Some pasted pupil work into free tools; others avoided AI altogether. Leaders could not say which tools were in use or what data had gone into them, and reports and lesson preparation were still eating teachers’ evenings.
The solution
Policy first, then the work, then the people.
We wrote the trust’s AI policy with its leaders, redesigned the two tasks that cost teachers the most time, and proved the change in four schools before it went to every school in the trust.
Policy
- Acceptable use for staff and pupils.What AI may be used for, by whom and with which tools, written in plain English for each audience.
- Assessment integrity.Rules for coursework and non-exam assessment in line with JCQ guidance, and a fair process when misuse is suspected.
- Data protection.An impact assessment for every approved tool, and no pupil data in any tool that learns from it.
- Safeguarding.AI written into the safeguarding policy, with filtering and monitoring checked against the DfE standards.
- Governance.An approved tools register, an owner for each tool and a termly review by the trust board.
Process
- Where the time goes.A workload study across four schools found report writing and lesson resourcing took the most hours.
- Report writing.Teachers draft from their own markbook notes, AI tidies the wording, and the teacher signs off every line.
- Lesson resources.A shared library of resources, with AI used to adapt them for reading age and special educational needs.
- Letters home.Templates and translation for parent communications, always checked by staff before they are sent.
- Measured, not assumed.Time logged before and after in the pilot schools; only what moved went trust-wide.
Change
- A pilot in four schools.One term, with a success measure and a stop condition agreed with the trust board.
- A champion in every school.A trained lead teacher in each school, with protected time to support colleagues.
- Training by role.Short sessions for teachers, support staff, leaders and governors, each on what changes for them.
- Families told plainly.An explainer for parents on how the schools use AI, and what they never use it for.
- Handover.Policy, training and the tools register owned by the trust’s own team, with a review calendar.
The payoff
Measured against the baseline.
- 4 hourssaved per teacher in each reporting cycle
- 9 in 10teachers using the approved tools by the end of the pilot
- 1 termfrom pilot to a trust-wide decision
- 100 %of reports signed off line by line by the teacher
Not everything moved. Lesson resourcing saved less time than reports: adapting a resource still needs a teacher’s judgement, so the trust kept that tool optional.
What’s next
Built to carry on without us.
The trust now reviews new tools against its register each term, and is extending the same approach to admissions and the support of pupils with special educational needs.
Talk to usEvery first conversation is with a partner, and costs nothing.
Ask us about AI in your schools