What problem does it solve?
Planning lessons and preparing resources is a large, steady part of a teacher's week. In the randomized trial of Oak National Academy's Aila, teachers who did not use it spent about three hours 19 minutes a week on lesson and resource preparation, and the evaluation found that almost all participating teachers regarded lesson planning as a source of stress and felt they spent too much time on it. The New South Wales government ties its work on the teacher shortage to a commitment to cut teachers' administrative work by five hours a week.
A general chatbot can draft a worksheet in seconds, but the draft has to fit the curriculum, the year group and the reading level of the class, and the teacher has to check it before pupils see it. The job of a lesson assistant is to produce a first draft that is close enough to the school's own scheme of work to be worth editing, and to cover the adjacent chores that eat time, such as differentiating a resource for pupils of different ability or checking a newsletter.
The evidence so far is modest and uneven. An independent randomized trial of Aila found a saving of around 49 minutes a week and resources that were no worse, with light use that declined over ten weeks. Education departments in New South Wales and Singapore have rolled out their own assistants to teachers, and we found no measured average saving published by either.
- The EEF commissioned evaluation of Oak National Academy's Aila, run by NFER, found that almost all participating teachers regarded lesson planning as a source of stress and felt they spent too much time on it, as reported by EdTech Innovation Hub.Oak's Aila saves primary teachers 49 minutes a week on lesson planning, EEF trial finds (2026)
- The New South Wales government says it is committed to reducing the admin load on teachers by five hours per week, and describes NSWEduChat as one tool aimed at routine administrative tasks.Expansion of AI education tool trial delivers more time for teachers in our classrooms (2024)
How does it work?
- The teacher states the goal. The teacher starts from a learning outcome, a year group and a subject, and can add what the class already knows or the context of the school.
- The assistant drafts from curriculum content. It builds sections, activities and components for a lesson, drawing on the curriculum aligned material the school or education body provides. Oak's Aila produces a lesson plan, a slide deck, two quizzes and a worksheet for each lesson; Singapore's Authoring Copilot proposes sections, activities and components in any subject and level.
- The teacher edits step by step. Content can be added or removed, the reading age lowered or the context changed, and the files stay fully editable.
- The assistant adapts what the teacher already has. A resource can be re levelled for pupils of different ability or adjusted to their interests, and a letter or newsletter can be checked. The Aila trial found that teachers often wanted to adapt existing material rather than start from scratch. Oak says it is now bringing Aila's capabilities into its existing lessons and resources, and NSWEduChat's teacher version is aimed at adjusting resources to different ability levels.
- The teacher uses the result in class. The draft is a starting point: the teacher checks it for accuracy, sequencing and suitability for the class, and the teacher decides what to teach and how to teach it.
- Audience
- Employee facing
- Autonomy
- Copilot
- Adoption
- Early adopters
- Channels
- Web chat
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Hours saved | Not pooled | about 0.8 hours | 1 | 1 independent |
| Productivity gain | Too few to pool | 24% | 1 | 1 independent |
Value drivers: Employee productivity, Speed and cycle time.
Indicative value
A primary school in England with 20 classroom teachers
GBP 4500 to GBP 25,480
Value of teacher planning time released per year
How this is calculated
Formula: teachers * minutesSaved / 60 * weeks * hourlyCost. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Classroom teachers who have the assistant teachers, teachers | 20 | 20 | The reference school. |
| Minutes of planning time saved per teacher per week minutesSaved, minutes per teacher per week | 15 | 49 | The high end is the average in the EEF commissioned Aila trial with primary teachers (around 49 minutes a week over ten weeks, reported by EdTech Innovation Hub). The low end allows for the light and declining use seen in that trial. Editorial assumption, replace with your own time diary data. Source |
| Teaching weeks per year weeks, weeks per year | 36 | 39 | Editorial assumption, replace with your school calendar. |
| Cost of one teacher hour hourlyCost, GBP per hour | 25 | 40 | Editorial assumption for a fully loaded teacher hour. Replace with your own figure. |
What it leaves out: The result prices time, not cash: nobody's pay falls and the hours only matter if teachers use them on pupils or on rest. It leaves out licence or build cost, training, the time teachers spend checking and correcting drafts, and any effect on lesson quality, which the trial measured only as "no worse". The trial covered primary teachers over one autumn term, so secondary schools and longer periods may differ.
Who already uses it?
3 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
NSW Department of Education
Australia · Education · 2024
The NSW Department of Education built NSWEduChat, a generative AI app with separate versions for teachers and students, hosted in the department's own cloud environment. The teacher version aims to cut time spent on routine administrative work: producing classroom resources at different ability levels, adjusting resources to students' interests, and checking correspondence and newsletters.
It started in 16 schools at the start of 2024, grew to 50 schools in Term 2 and was announced for all public school teachers from the start of Term 4, with optional use. A September 2025 department release says the teacher version had been rolled out to all schools.
The department reports positive trial feedback, including that some teachers saved over an hour a week producing lesson resources, and later said surveys show the teacher version saves time in producing resources for different ability levels. It has not published a measured average.
No outcome disclosed.
Oak National Academy
United Kingdom · Education · 2025
Oak National Academy, which publishes free curriculum resources for teachers in England and which the Department for Education says it funded to develop AI tools for teachers, launched its AI lesson assistant Aila in September 2024. A teacher states what to teach and works step by step with Aila, which draws on Oak's curriculum aligned content to produce an editable lesson plan, slide deck, quizzes and worksheet.
The Education Endowment Foundation commissioned the National Foundation for Educational Research to run a ten week randomized trial in autumn 2025 with teachers of Years 3 to 6 in England. Teachers assigned to Aila spent around 49 minutes less per week on lesson and resource preparation than the comparison group, and independent assessors judged their resources no worse.
Use was light: 81% of the Aila group opened it at least once, but it supported only around 7% of lessons in a typical week and use declined over the ten weeks. Gains were largest for early career teachers, and trade press reports that Oak's chief executive said Oak's own earlier surveys had suggested bigger savings than the trial found.
- Hours saved: about 0.8 hours, per teacher per week
"Oak National Academy’s AI lesson assistant Aila saved primary teachers an average of around 49 minutes a week on lesson and resource preparation without reducing the quality of the materials they produced, according to a new independent randomized controlled trial."
Claimed by: independent - Productivity gain: 24%
"Primary teachers who used Aila, the AI lesson assistant from Oak National Academy, spent 24% less time preparing lessons, with no drop in the quality of what they produced, according to a randomised trial whose results were published on Tuesday 6 October."
Claimed by: independent
Ministry of Education, Singapore
Singapore · Education · 2024
Singapore's Ministry of Education rolled out the Authoring Copilot in June 2024 on the Student Learning Space, the national online learning portal for all MOE schools. A teacher gives the tool an input and it proposes sections, activities and components for a lesson in any subject and level, generating draft lesson ideas and learning resources that the teacher adapts. A second teacher tool on the same portal, the Data Assistant, analyses students' written responses to surface common themes and misconceptions.
GovTech Singapore describes both as built to augment teachers, with human in the loop or human over the loop moderation. MOE has published no figure for teacher time saved. One primary school teacher quoted by The Straits Times says she saves a lot of time in lesson planning with it, and notes that the tool could not yet read Chinese, so she translates her inputs into English.
No outcome disclosed.
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- The curriculum, schemes of work and exemplar lessons the assistant should draw on, with the right to use them
- School policies that affect resources, such as reading age, accessibility and safeguarding wording
- A baseline of how long teachers spend on planning, from a short diary or survey
Systems to integrate
- School or district identity provider for single sign on
- Document storage or learning platform where teachers keep and share resources
- Staff messaging channel such as Microsoft Teams, if teachers should reach it there
Complexity: Low
A planning assistant needs the school's or the education body's curriculum documents, staff sign in through the school's identity provider and a data protection impact assessment for staff accounts. No pupil data, payment or core system is involved when it is used for planning.
- 1
Start with the teachers most likely to gain
The Aila trial found the largest time savings among early career teachers, those less confident in their subject or teaching knowledge and those comfortable with technology. Offer the tool, with training, to those groups first and ask other teachers to opt in.
- 2
Ground the assistant in your curriculum
Index the scheme of work, exemplar lessons and approved resources, so drafts use the school's sequence, vocabulary and examples, and make the assistant say when the curriculum does not cover a request.
- 3
Support adapting existing material as well as planning from scratch
In the Aila trial teachers often adapted material they already had, and the evaluation links the low share of lessons planned with the tool (about 7% in a typical week) to a design built around lessons from scratch. Let teachers paste in or upload a resource and ask for a differentiated version, a shorter version or a quiz.
- 4
Agree what teachers must check
The EEF cautions that saving time does not automatically mean better teaching and that teachers still need to check AI drafts for accuracy, sequencing and suitability. Give them a short checklist and make clear that the teacher owns every resource.
- 5
Keep pupil information out
Planning needs no names or personal data. Block pupil identifiers from prompts and tell teachers not to paste in reports or assessment data, unless the school has assessed that use.
- 6
Measure against a baseline for a full term
Oak's own earlier surveys suggested bigger savings than the independent trial found, so use a time diary against a baseline or comparison group. Track weekly use as well, because use declined over the ten weeks of the trial.
Guardrails
- Staff accounts only, through the school's single sign on
- No pupil names or personal data in prompts, with masking where staff paste in text
- Drafts grounded in approved curriculum content, with a clear message when it does not cover a request
- Age and reading level settings, and content filters suitable for school
- A visible AI label on every draft until a teacher has reviewed it
KPIs to instrument
- Minutes spent on planning and resource preparation per teacher per week, against the baseline
- Weekly active teachers and the share of lessons planned with the assistant, tracked for the whole term
- Share of drafts edited before use and the kinds of edits teachers make
- Quality of resources on a blind sample reviewed by subject leads
- Teacher views on workload and on whether the tool changed how they plan
Human in the loop
The teacher is the author of every resource and checks it before it reaches pupils. Subject leads review a sample of AI drafts each term for accuracy and curriculum fit, and the school decides which uses are out of scope, such as writing reports or marking pupils' work.
Common failure modes
- Light use that fades
- The trial saw 81% of the Aila group open the tool but only about 7% of lessons planned with it in a typical week, with use falling over ten weeks. Put the tool inside the resources teachers already use and support adaptation.
- Plausible errors reach pupils
- A fluent draft can contain a wrong fact, an out of sequence step or an unsuitable example. Ground drafts in the curriculum, require a teacher check and sample resources for accuracy.
- Time saved with nothing gained
- The EEF notes that saving planning time does not by itself improve teaching. Agree up front where released time should go, such as feedback or small group work, and review that too.
- Language and subject gaps
- One Singapore teacher reported that the Authoring Copilot could not yet read Chinese, so she translated her inputs into English. Test the assistant in every subject and language the school teaches before rollout.
What are the risks and rules?
EU AI Act
Minimal risk
A tool that only drafts planning materials for teachers is not listed in Annex III. Annex III point 3 covers education systems that determine access or admission, evaluate learning outcomes, assess the level of education a person will receive or monitor prohibited behaviour in tests. The assistant would become high risk under point 3(b) if it were used to evaluate pupils' work or steer a pupil's learning on the basis of its outcomes. The AI literacy duty in Article 4 applies to schools as deployers.
Guidance
- Guidance for generative AI in education and research (UNESCO, Global). UNESCO's global guidance on generative AI in education; it proposes mandating the protection of data privacy and setting an age limit for independent conversations with GenAI platforms.
- Generative artificial intelligence (AI) in education (UK Department for Education, Europe). The department's position on generative AI in schools and colleges; it states that it has funded Oak National Academy to develop AI tools that speed up lesson planning and reduce workloads.
- Article 4, AI literacy (European Union, Europe). Providers and deployers of AI systems must take measures to support the AI literacy of their staff and of other persons dealing with the systems on their behalf.
- Annex III, high risk AI systems (point 3, education and vocational training) (European Union, Europe). Lists AI used to determine access or admission, to evaluate learning outcomes, and to assess the level of education a person will receive as high risk, which marks the line a planning assistant should not cross.
Controls to put in place
- A data protection impact assessment for staff accounts, covering prompts and logs
- An approved tools list and a school policy on what teachers may and may not put into the assistant
- Staff training on AI use and on checking drafts, as part of the school's AI literacy measures
- Inventory entry with an accountable owner and a review date
- A termly sample check of drafts for accuracy and curriculum fit
Frequently asked questions
- How much time does an AI lesson assistant save teachers?
An independent randomized trial found that primary teachers in England who were assigned to Oak's Aila spent around 49 minutes a week less on lesson and resource preparation than a comparison group, with resources judged no worse. The tool was used for only about 7% of lessons in a typical week and use declined over ten weeks. The NSW government says some teachers saved over an hour a week with NSWEduChat, but that is a report from some teachers and not an average.
- Which teachers benefit most?
In the Aila trial, gains were largest for early career teachers, for teachers less confident in their subject or teaching knowledge and for teachers comfortable with technology. Fewer than one in five teachers said the tool had meaningfully changed how they taught or planned, and many preferred other AI tools.
- Is an AI lesson planning assistant high risk under the EU AI Act?
Usually not. Drafting planning materials for teachers is not an Annex III use. It would become high risk under Annex III point 3(b) if it were used to evaluate pupils' work or to steer a pupil's learning on the basis of outcomes, so keep marking and placement decisions out of the assistant. Schools still need to meet the AI literacy duty in Article 4.
How to cite this page
Blits.ai AI Use Case Library, "AI assistant for teachers: lesson planning and classroom resources", last verified 13 October 2026, https://www.blits.ai/ai-use-cases/teacher-lesson-planning-and-resource-assistant. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 13 October 2026: First published