AI for schools and trusts

We run an AI Audit of where AI fits in a school or trust. You get a findings report, a prioritised roadmap and a costed business case. What follows is built on the school’s own mark schemes, and staff are trained to use it inside Classroom or Teams.

No, and the DfE product safety standards require AI tools to mitigate cognitive deskilling. The workflows that earn their place are marking first passes, lesson planning support, parent communications and admin triage. The EEF study found teachers saved around 31% of planning time and still owned every plan.

The DfE turned its expectations into formal product safety standards on 19 January 2026, covering content filtering, distress detection, cognitive deskilling, manipulation defences and a ban on training with learner or teacher work. We design against the standards, document the alignment, and hand it to your DSL and DPO.

JCQ treats undeclared AI use in submitted work as malpractice, and recorded 1,125 GCSE and A-level disqualifications in 2025. Anything we build that touches assessed work keeps the student declaration trail intact, the teacher in the moderation loop and the evidence defensible at a JCQ inquiry.

Yes. The default is an Azure UK tenant or an equivalent UK or EU region, and on-prem is supported where sovereignty demands it. Models run behind the trust’s own identity provider, logs are kept for moderation rather than training, and the trust can export and delete everything.

Google Classroom and Microsoft Teams for Education are evidenced production integrations: Rubrical syncs assignments and returns grades through both. Microsoft 365 and Entra ID are standard. SIMS, Arbor or Bromcom integration is scoped per engagement, so tell us which one you run at the audit.

The EU AI Act classifies systems that evaluate learning outcomes or monitor candidates during tests as high-risk. The stand-alone compliance deadline moved to December 2027, but the classification is settled and the institution carries its own duties, including a Fundamental Rights Impact Assessment. We write that alongside the build. If you want to check which duties reach you first, there is a free <a href="https://aigovernancecheck.co.uk/eu-ai-act-check/">EU AI Act applicability check</a> at AI Governance Check.

Most engagements open with the AI Audit, mapping where AI fits and where it should not go. Scope, fee and schedule are agreed in your proposal. Rubrical, our DfE partnership, ran from discovery to a fourteen-school rollout, and a build is scoped against what the audit surfaces.

The audit is £10,000 excluding VAT at standard scope, agreed before the work starts. An AI Charter can be added for a further £3,000 excluding VAT, scope agreed separately. An Embedded AI Lead engagement is from £5,000 a month excluding VAT. We give you indicative figures on a call.

Those are products you subscribe to. We start with an assessment, and the audit chooses the workflow. A build makes it work inside your stack, and we either hand over or continue as an Embedded AI Lead. We run a DfE programme-partner build in production and hold three certifications.

Faculty, now part of Accenture, operates at enterprise scale on DfE central programmes and Big Four day rates. We work at trust scale: independent schools, multi-academy trusts of three to thirty schools, colleges and university departments. The entry point is small enough to sign off without a procurement exercise.