Every document, one account
Lesson plans, worksheets, quizzes, presentations, IEPs, scope-and-sequence documents (the plan of what a course covers, in what order) and student reports. Voice and audio features help with accessibility and onboarding.
EdTech AI case study: K-12 schools
An AI teaching platform that saves each teacher 3.6 hours a week.
An EdTech company wanted teachers to stop losing evenings and weekends to planning and paperwork, using AI that schools would actually approve. We designed and built it, from the first curriculum-mapping prototype to a live product schools can buy.
- Localized into 108 languages
- Passed an independent school-safety assessment
- Student data never trains the AI models
- Generative AI
- Curriculum alignment
- Astro
- GDPR
In brief
- The client
- An EdTech company focused on teacher workload, serving K-12 schools in Australia and New Zealand and expanding internationally.
- The problem
- Planning and paperwork ate into teachers' evenings. Chatbots didn't know the curriculum, and schools couldn't trust them with student data.
- The result
- A live platform used from single classrooms to multi-school rollouts. Each active teacher saves 3.6 hours a week on average.
The challenge
Teachers were spending their weekends on paperwork.
Teaching doesn't stop when the bell goes. Lesson plans, worksheets, marking and report writing all sit on top of classroom time. The client's early user research confirmed hours of unpaid weekend work on content that AI could speed up, if schools could trust it.
General-purpose chatbots don't fix this well. They don't know the national or regional curriculum a school teaches to. They make no promise about what happens to student data. And school leaders have no way to see how staff are using them.
So the client wanted a platform built for the job. For one teacher, it had to produce curriculum-aligned lesson plans, worksheets, assessments and reports in minutes. For a whole school, it had to be something leadership could govern.
What the platform had to do
- Cover every document in one tool. Teachers needed dozens of output types, from a quick quiz to a formal Individual Education Plan (IEP) for a student with additional needs, without juggling a separate tool for each.
- Match the curriculum. Content had to line up with specific national and regional curriculum standards. Generic material that a teacher still has to rework saves nobody any time.
- Fit each learner. Differentiated instruction meant adapting output to reading level, learner needs and regional spelling or dialect, including switching between variants of English.
- Give school leaders control. Leaders needed shared templates, usage reporting and moderation across departments. Without that, there was no case for a school-wide purchase.
- Keep student data safe. Student data could not be used to train the underlying AI models. Many schools also wouldn't trial the product at all until it had passed an independent school-safety assessment.
What we built
Two kinds of buyer, served from one codebase.
We built a multi-tool generative AI content platform for individual teachers and whole-school administrators, on a single shared codebase (one set of code to build and maintain). Our work covered product design, generative AI engineering and the compliance work schools require.
Aligned to the curriculum
A curriculum-alignment layer maps what the AI writes to configurable national and regional standards, so teachers rework far less of it.
In 108 languages
Localization into 108 languages, with the dialect and spelling differences of regional English, and bilingual output.
A tier for school leaders
Usage analytics, shared templates and accounts set up per teacher, which supports volume licensing and department-wide rollouts. Each school defines its approved slide and document layouts, and the AI output follows its branding and formatting automatically.
Ready for school procurement
Encryption at rest and in transit (when data is stored and when it is sent), GDPR compliance and a strict rule against training models on student data, validated through an independent school-safety assessment program.
Easy to find, measured from the start
The marketing and product front end is built in Astro, a framework for fast pages that search engines read well, so a teacher who finds one tool reaches a trial quickly. Adoption analytics track hours saved, workflows generated and weekly active use, giving the client measured outcomes for sales, marketing and renewals.
The results
From one classroom to multi-school rollouts.
The platform's usage data backs up the time savings the client set out to deliver. The figures are the client's own, as reported to us.
- 20,000+
- hours of teacher time saved to date, and rising
- 32,000+
- teaching workflows generated, from lesson plans and worksheets to assessments, presentations and reports
- 3.6 hrs
- saved per active teacher, per week, on average
- 108
- languages supported
What made it work
In schools, procurement gets a vote before teachers do.
The product had to win teachers and pass procurement at the same time.
Schools are a hard market for anything new. Procurement moves slowly and asks pointed questions about data. Teachers drop a tool the moment it creates more work than it saves. And a teacher can love a tool and still not be allowed to use it.
Safety questions were product requirements
We treated the independent safety assessment and the rule that student data never trains a model as product requirements from the first build. A no-training promise added after launch means explaining to every school why it wasn't there before, and that is a hard conversation to win.
Teachers adopt it, schools buy it
- Tiered pricing. Free, individual paid, and per-teacher volume licensing for schools, so teachers can adopt it bottom-up while schools buy it top-down.
- Grounds to approve. Independent assessment for safe use in schools gives procurement teams the grounds to approve school-wide and district-wide rollouts.
- Across school settings. It is in use across primary, secondary and special-education settings, from teachers who signed up on their own to whole-school and multi-school licenses.
Why the client chose us
We understood the trade-off between teachers and procurement going in. We had already shipped products in regulated, safety-sensitive categories. We knew how to give teachers the generative AI they wanted without cutting corners on the data governance schools require, and how to make one product work for a teacher who wants speed and an administrator who wants oversight.
Related services
Services for a project like this.
Start from the question closest to yours.
- We want AI that writes material to our customers' own standards and formats.Generative AI development
- Our buyers ask how we handle personal data before they will sign.GDPR compliance
- We want our AI features checked before a buyer's security review.AI security
- We need a fast web product and a site that people can find.Web and custom software
More case studies: the AI recruitment platform, the patient care management app, the ISO 27001 certification program and the bank security assessment. Or see every case study.
Let's talk
Selling AI into a market that asks hard questions about data?
Schools, clinics and banks all want to know where the data goes. We design around that answer early. Tell us about your product and who buys it. We reply within one working day.