TL;DR:

• The Department for Education has opened a tender inviting EdTech firms and AI labs to build curriculum-aligned tutoring tools for Years 9 and 10 pupils across English, maths, science and modern foreign languages.
• Up to eight companies will begin school testing this summer, with successful tools rolled out nationally from 2027 — potentially reaching 450,000 disadvantaged pupils a year.
• The procurement builds on the government’s “Every Child Achieving and Thriving” white paper and puts UK DfE among the most concrete state-commissioned AI education programmes anywhere.

The DfE’s requirement is that any product align to the national curriculum, integrate with classroom practice under teacher supervision, and demonstrably serve disadvantaged pupils. The brief explicitly references evidence that private tutoring can accelerate learning by up to five months — and that its cost puts it out of reach of most families.

Why this procurement is different

Most national AI-in-education announcements have been frameworks, advisory notes or voluntary pilot funding. This is an actual government tender with school-based testing from this summer and a 2027 national target, using procurement — not grants — as the policy instrument. The DfE is signalling that adaptive, one-to-one-style support should no longer be privately rationed; it wants the private-tutoring benefit delivered inside the state system via AI.

The safety and evidence bar

The tender puts safety at the front. Every bidder must show accessibility for pupils with different needs, how the tool will not displace teacher judgement, and how it will be tested in real classrooms before scale. The DfE’s framing — as an “additional layer of assistance” that frees teachers to focus on pupils most at risk of falling behind — tries to sidestep the automation-displacement argument that has dogged earlier EdTech rollouts. Whether that framing survives the first set of classroom results will depend on teacher feedback more than vendor benchmarks.

UK context

The tender sits alongside Ofqual’s same-week comments about potentially scrapping extended-writing coursework because of AI-cheating risk. Both items point to the same structural question: where AI tools should sit in the assessment and learning pipeline. The DfE is trying to pull AI into the learning stage, where it helps; Ofqual is working out how to keep it out of the assessment stage, where it distorts.

Looking forward

The first commercial signal will be the shortlist of up to eight suppliers this summer. Expect bids from UK EdTech incumbents, US platform giants with local footprint, and at least one Sovereign AI-backed startup — the government has stated it wants British AI companies solving British problems, and this is the most obvious first test of that doctrine in a live tender.