In one trial, an AI assistant cut primary teachers' lesson planning by about 49 minutes a week

We test whether an AI tool saves your staff more time than its licence and checking cost, before you sign a contract. Then we map it to UK GDPR and, for schools, to the ICO's Children's code.

Every figure sourced and datedFounded 2025, no education clients yet

ivory and oat paper layers joined by one teal thread

In a randomised trial of 464 primary teachers in 108 English schools, some were given Oak National Academy's AI assistant, Aila [1]. They spent 2.5 hours a week preparing lessons, against 3.3 hours for teachers without it [1]. That is about 49 minutes a week saved, with resource quality no worse [1]. New teachers and those less sure of their subject were more likely to save time [1].
The trial ran for ten weeks in autumn 2025, and teachers used Aila for only 7 per cent of lessons [1]. Use fell over the trial, and both groups could still use ChatGPT [1]. We baseline planning minutes first and stop any tool that does not move them.
Sources
1. Palak Roy and others, Lesson planning using AI lesson assistant, Aila: a Teacher Choices trial. Evaluation report (Education Endowment Foundation, 2026), p. 6 <https://d2tic4wvo1iusb.cloudfront.net/production/documents/projects/aila-lesson-planning_teacher-choices_evaluation-report.pdf> [accessed 31 October 2026]
A June 2026 Department for Education study tracked which teachers left state schools. Hours worked did not predict leaving, but rating the workload as less acceptable did, even at equal hours [1]. Weak support from leaders predicted leaving more strongly, and once it was counted, workload lost its consistent link [1]. The authors say the links are correlational [1].
In a June 2026 YouGov poll for Bett and Lenovo (vendor-reported), 51 per cent of teachers who use AI said it eased their workload [2][3]. Only 35 per cent worked fewer hours [3]. Ten unions want staff to be partners when schools buy AI [4]. We agree with staff and union reps where freed time goes, then track staff ratings of workload and of leadership support each term.
Sources
1. UCL Institute of Education, Working lives of teachers and leaders: teachers and leaders leaving state education, waves 1 to 3 (Department for Education, 2026), pp. 27, 43, 50 <https://www.gov.uk/government/publications/teachers-and-leaders-leaving-english-state-education-waves-1-to-3> [accessed 31 October 2026]
2. Bett Blog Editorial Team, 'AI use reaches 80% among teachers in the United Kingdom, but confidence in the technology is waning', Bett Brasil, 8 September 2026 <https://brasil.bettshow.com/en/bett-blog/uso-de-ia-chega-80-entre-professores-reino-unido-mas-confian-na-tecnologia-perde-> [accessed 31 October 2026]
3. Cerys Turner, 'Teachers' use of AI soars - but has little impact on working hours', Tes, 27 August 2026 <https://www.tes.com/magazine/news/general/teacher-use-ai-soars-little-impact-working-hours> [accessed 31 October 2026]
4. Trades Union Congress, 'An urgent call for educator voice in AI and EdTech from 10 unions', TUC, 16 January 2026 <https://www.tuc.org.uk/urgent-call-educator-voice-ai-and-edtech-10-unions> (archived at <https://web.archive.org/web/20260301150544/https://www.tuc.org.uk/urgent-call-educator-voice-ai-and-edtech-10-unions>) [accessed 31 October 2026]
A June 2026 report for the Department for Education found that suppliers quote total user numbers at renewal, which do not separate quick logins from steady use [1]. So schools risk renewing tools that staff rarely use [1]. For universities, a vendor-funded 2026 US poll of 130 campus technology chiefs found 29 per cent said AI spending met or beat expectations [2][3].
We size each use before you buy. If the Aila trial's 49 minutes held for all 38 teaching weeks, that is about 31 hours per teacher [4][5]. At the top main-scale salary, £46,940 over 1,265 directed hours, it is worth about £1,140 [5]. That is a ceiling, and time rather than cash. A tool whose licence and checking time cost more is marked do not buy.
Sources
1. Martina Bolibruchova and others, Assessment of the education technology market in England: research report (Department for Education, 2026), p. 59 <https://www.gov.uk/government/publications/assessment-of-the-education-technology-market-in-england> [accessed 31 October 2026]
2. Colleen Flaherty and Sara Custer (eds), 2026 Survey of Campus Chief Technology/Information Officers (Inside Higher Ed and Hanover Research, 2026), pp. 4-5, 19, 57 <https://www.secureworld.io/hubfs/documents/2026-IHE-CTO-CIO-Survey_Final__0.pdf> [accessed 31 October 2026]
3. Inside Higher Ed, '2026 Survey of Campus Chief Technology/Information Officers', Inside Higher Ed, 11 May 2026 <https://www.insidehighered.com/reports/2026/05/11/2026-survey-campus-chief-technologyinformation-officers> [accessed 31 October 2026]
4. Palak Roy and others, Lesson planning using AI lesson assistant, Aila: a Teacher Choices trial. Evaluation report (Education Endowment Foundation, 2026), p. 6 <https://d2tic4wvo1iusb.cloudfront.net/production/documents/projects/aila-lesson-planning_teacher-choices_evaluation-report.pdf> [accessed 31 October 2026]
5. Department for Education, School teachers' pay and conditions document 2026 and guidance on school teachers' pay and conditions (Department for Education, 2026), pp. 21, 48-49 <https://www.gov.uk/government/publications/school-teachers-pay-and-conditions> [accessed 31 October 2026]
In a late-2025 government survey, 49 per cent of primary schools and 73 per cent of secondaries found a breach or attack, most often phishing [1]. So did 88 per cent of 33 colleges and 98 per cent of 49 universities, small samples with wide margins [1]. Of colleges and universities hit, 49 per cent reported a negative outcome [1].
In June 2026 the ICO reported on its 2024 and 2025 audits of 28 edtech suppliers, which left out generative AI assistants [2]. Many reused children's data without recognising they were controllers, and some used it to train AI features [2]. Those acting beyond a school's instructions fall under the Children's code [2]. We map who controls each use, then check contracts and DPIAs before AI reads pupil records.
Sources
1. Department for Science, Innovation and Technology and Home Office, 'Cyber security breaches survey 2025/2026: education institutions findings', GOV.UK, 30 April 2026 <https://www.gov.uk/government/statistics/cyber-security-breaches-survey-20252026/cyber-security-breaches-survey-20252026-education-institutions-findings> [accessed 31 October 2026]
2. Information Commissioner's Office, Edtech examined: key findings from our audits (ICO, 2026), pp. 2, 9, 11, 18, 67 <https://ico.org.uk/action-weve-taken/audits-and-overview-reports/2026/06/edtech/> [accessed 31 October 2026]
The EU AI Act of 2024 lists AI used to decide admission or to evaluate learning as high-risk [1]. As amended in 2026, those duties apply from 2 December 2027 [2]. They can reach UK institutions where the AI's output is used in the EU [3].
UK law applies now. Since 5 February 2026, a significant, solely automated decision about a student needs safeguards, including human review and a way to contest it [4]. In June 2025, its newest statement we found, the Office for Students called its approach to AI principles-based [5]. The DfE's product safety standards for England, updated in January 2026, expect school AI suppliers to follow the Children's code [6]. Your DPO gets one register per AI use, naming the rule and who holds the evidence.
Sources
1. European Commission, 'Annex III: High-risk AI systems referred to in Article 6(2)', AI Act Service Desk, Regulation (EU) 2024/1689 (2024) <https://ai-act-service-desk.ec.europa.eu/en/ai-act/annex-3> [accessed 31 October 2026]
2. European Commission, 'Article 113: Entry into force and application', AI Act Service Desk, consolidated text as at 27 July 2026 <https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-113> [accessed 31 October 2026]
3. European Commission, 'Article 2: Scope', AI Act Service Desk, consolidated text as at 27 July 2026 <https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2> [accessed 31 October 2026]
4. UK General Data Protection Regulation, art. 22C, 'Safeguards for automated decision-making', legislation.gov.uk, inserted by the Data (Use and Access) Act 2025, s. 80, in force 5 February 2026 <https://www.legislation.gov.uk/eur/2016/679/article/22C> [accessed 31 October 2026]
5. Josh Fleming, 'Embracing innovation in higher education: our approach to artificial intelligence', Office for Students, 5 June 2025 <https://www.officeforstudents.org.uk/news-blog-and-events/blog/embracing-innovation-in-higher-education-our-approach-to-artificial-intelligence/> [accessed 31 October 2026]
6. Department for Education, 'Generative AI: product safety standards', GOV.UK, updated 19 January 2026 <https://www.gov.uk/government/publications/generative-ai-product-safety-standards/generative-ai-product-safety-standards> [accessed 31 October 2026]

In one trial, an AI assistant cut primary teachers' lesson planning by about 49 minutes a week

We test whether an AI tool saves your staff more time than its licence and checking cost, before you sign a contract. Then we map it to UK GDPR and, for schools, to the ICO's Children's code.

Book a call with Khaled

Khaled Shivji

Founder and CEO, Principal AI Consultant

In a randomised trial of 464 primary teachers in 108 English schools, some were given Oak National Academy's AI assistant, Aila [1]. They spent 2.5 hours a week preparing lessons, against 3.3 hours for teachers without it [1]. That is about 49 minutes a week saved, with resource quality no worse [1]. New teachers and those less sure of their subject were more likely to save time [1].
The trial ran for ten weeks in autumn 2025, and teachers used Aila for only 7 per cent of lessons [1]. Use fell over the trial, and both groups could still use ChatGPT [1]. We baseline planning minutes first and stop any tool that does not move them.
Sources
1. Palak Roy and others, Lesson planning using AI lesson assistant, Aila: a Teacher Choices trial. Evaluation report (Education Endowment Foundation, 2026), p. 6 <https://d2tic4wvo1iusb.cloudfront.net/production/documents/projects/aila-lesson-planning_teacher-choices_evaluation-report.pdf> [accessed 31 October 2026]
A June 2026 Department for Education study tracked which teachers left state schools. Hours worked did not predict leaving, but rating the workload as less acceptable did, even at equal hours [1]. Weak support from leaders predicted leaving more strongly, and once it was counted, workload lost its consistent link [1]. The authors say the links are correlational [1].
In a June 2026 YouGov poll for Bett and Lenovo (vendor-reported), 51 per cent of teachers who use AI said it eased their workload [2][3]. Only 35 per cent worked fewer hours [3]. Ten unions want staff to be partners when schools buy AI [4]. We agree with staff and union reps where freed time goes, then track staff ratings of workload and of leadership support each term.
Sources
1. UCL Institute of Education, Working lives of teachers and leaders: teachers and leaders leaving state education, waves 1 to 3 (Department for Education, 2026), pp. 27, 43, 50 <https://www.gov.uk/government/publications/teachers-and-leaders-leaving-english-state-education-waves-1-to-3> [accessed 31 October 2026]
2. Bett Blog Editorial Team, 'AI use reaches 80% among teachers in the United Kingdom, but confidence in the technology is waning', Bett Brasil, 8 September 2026 <https://brasil.bettshow.com/en/bett-blog/uso-de-ia-chega-80-entre-professores-reino-unido-mas-confian-na-tecnologia-perde-> [accessed 31 October 2026]
3. Cerys Turner, 'Teachers' use of AI soars - but has little impact on working hours', Tes, 27 August 2026 <https://www.tes.com/magazine/news/general/teacher-use-ai-soars-little-impact-working-hours> [accessed 31 October 2026]
4. Trades Union Congress, 'An urgent call for educator voice in AI and EdTech from 10 unions', TUC, 16 January 2026 <https://www.tuc.org.uk/urgent-call-educator-voice-ai-and-edtech-10-unions> (archived at <https://web.archive.org/web/20260301150544/https://www.tuc.org.uk/urgent-call-educator-voice-ai-and-edtech-10-unions>) [accessed 31 October 2026]
A June 2026 report for the Department for Education found that suppliers quote total user numbers at renewal, which do not separate quick logins from steady use [1]. So schools risk renewing tools that staff rarely use [1]. For universities, a vendor-funded 2026 US poll of 130 campus technology chiefs found 29 per cent said AI spending met or beat expectations [2][3].
We size each use before you buy. If the Aila trial's 49 minutes held for all 38 teaching weeks, that is about 31 hours per teacher [4][5]. At the top main-scale salary, £46,940 over 1,265 directed hours, it is worth about £1,140 [5]. That is a ceiling, and time rather than cash. A tool whose licence and checking time cost more is marked do not buy.
Sources
1. Martina Bolibruchova and others, Assessment of the education technology market in England: research report (Department for Education, 2026), p. 59 <https://www.gov.uk/government/publications/assessment-of-the-education-technology-market-in-england> [accessed 31 October 2026]
2. Colleen Flaherty and Sara Custer (eds), 2026 Survey of Campus Chief Technology/Information Officers (Inside Higher Ed and Hanover Research, 2026), pp. 4-5, 19, 57 <https://www.secureworld.io/hubfs/documents/2026-IHE-CTO-CIO-Survey_Final__0.pdf> [accessed 31 October 2026]
3. Inside Higher Ed, '2026 Survey of Campus Chief Technology/Information Officers', Inside Higher Ed, 11 May 2026 <https://www.insidehighered.com/reports/2026/05/11/2026-survey-campus-chief-technologyinformation-officers> [accessed 31 October 2026]
4. Palak Roy and others, Lesson planning using AI lesson assistant, Aila: a Teacher Choices trial. Evaluation report (Education Endowment Foundation, 2026), p. 6 <https://d2tic4wvo1iusb.cloudfront.net/production/documents/projects/aila-lesson-planning_teacher-choices_evaluation-report.pdf> [accessed 31 October 2026]
5. Department for Education, School teachers' pay and conditions document 2026 and guidance on school teachers' pay and conditions (Department for Education, 2026), pp. 21, 48-49 <https://www.gov.uk/government/publications/school-teachers-pay-and-conditions> [accessed 31 October 2026]
In a late-2025 government survey, 49 per cent of primary schools and 73 per cent of secondaries found a breach or attack, most often phishing [1]. So did 88 per cent of 33 colleges and 98 per cent of 49 universities, small samples with wide margins [1]. Of colleges and universities hit, 49 per cent reported a negative outcome [1].
In June 2026 the ICO reported on its 2024 and 2025 audits of 28 edtech suppliers, which left out generative AI assistants [2]. Many reused children's data without recognising they were controllers, and some used it to train AI features [2]. Those acting beyond a school's instructions fall under the Children's code [2]. We map who controls each use, then check contracts and DPIAs before AI reads pupil records.
Sources
1. Department for Science, Innovation and Technology and Home Office, 'Cyber security breaches survey 2025/2026: education institutions findings', GOV.UK, 30 April 2026 <https://www.gov.uk/government/statistics/cyber-security-breaches-survey-20252026/cyber-security-breaches-survey-20252026-education-institutions-findings> [accessed 31 October 2026]
2. Information Commissioner's Office, Edtech examined: key findings from our audits (ICO, 2026), pp. 2, 9, 11, 18, 67 <https://ico.org.uk/action-weve-taken/audits-and-overview-reports/2026/06/edtech/> [accessed 31 October 2026]
The EU AI Act of 2024 lists AI used to decide admission or to evaluate learning as high-risk [1]. As amended in 2026, those duties apply from 2 December 2027 [2]. They can reach UK institutions where the AI's output is used in the EU [3].
UK law applies now. Since 5 February 2026, a significant, solely automated decision about a student needs safeguards, including human review and a way to contest it [4]. In June 2025, its newest statement we found, the Office for Students called its approach to AI principles-based [5]. The DfE's product safety standards for England, updated in January 2026, expect school AI suppliers to follow the Children's code [6]. Your DPO gets one register per AI use, naming the rule and who holds the evidence.
Sources
1. European Commission, 'Annex III: High-risk AI systems referred to in Article 6(2)', AI Act Service Desk, Regulation (EU) 2024/1689 (2024) <https://ai-act-service-desk.ec.europa.eu/en/ai-act/annex-3> [accessed 31 October 2026]
2. European Commission, 'Article 113: Entry into force and application', AI Act Service Desk, consolidated text as at 27 July 2026 <https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-113> [accessed 31 October 2026]
3. European Commission, 'Article 2: Scope', AI Act Service Desk, consolidated text as at 27 July 2026 <https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2> [accessed 31 October 2026]
4. UK General Data Protection Regulation, art. 22C, 'Safeguards for automated decision-making', legislation.gov.uk, inserted by the Data (Use and Access) Act 2025, s. 80, in force 5 February 2026 <https://www.legislation.gov.uk/eur/2016/679/article/22C> [accessed 31 October 2026]
5. Josh Fleming, 'Embracing innovation in higher education: our approach to artificial intelligence', Office for Students, 5 June 2025 <https://www.officeforstudents.org.uk/news-blog-and-events/blog/embracing-innovation-in-higher-education-our-approach-to-artificial-intelligence/> [accessed 31 October 2026]
6. Department for Education, 'Generative AI: product safety standards', GOV.UK, updated 19 January 2026 <https://www.gov.uk/government/publications/generative-ai-product-safety-standards/generative-ai-product-safety-standards> [accessed 31 October 2026]

Book a call with Khaled