We put AI right inside Word, PowerPoint, Teams, and Outlook, the tools you already use. Our team of experts connects it with every system you use, and work with you to configure the agents to your policies, and get the results you demand.
Teachers who use AI every week say it saves them about 5.9 hours a week. Most of that time comes from lesson content.
Only 19% of those teachers worked in a school with an AI policy. The rest used it on their own.
UAE teachers spend 5.4 hours a week marking and 3 hours on admin. That work hurts well-being more than teaching does.
Our reading AI use in schools is widespread, but personal, and it happens in a chat window. The documents an inspector reads have barely been touched.
The crunch points, from the school's side of the desk.
Swipe to see the whole year →
A made-up example, built the way a real one would be.
Tap the screenshot to open it full size
Try it yourself. You can do this by hand tomorrow. Next to every claim about teaching, write the outcome figure it should have moved, even when that figure went the wrong way.
The same judgement, written three ways.
1 Assertion
Teaching in Year 7 maths is good.
The inspector asks “How do you know?”
2 Assertion with data
Teaching in Year 7 maths is good: 74% of students met their MAP growth target this term.
The inspector asks “Is that growth down to teaching? What did you see in lessons?”
3 Triangulated judgement
Teaching in Year 7 maths is good: 74% met their MAP growth target, 5 of 6 observed lessons were good or better, and the October book look confirmed feedback is acted on.
The inspector asks “Can I see one of those lessons?”
Each item is dated and sourced. We update this list when a regulator does.
18 Sep 2026
Covers 170+ private schools and six safeguard areas. Training at three levels for 19,000 educators: Level 1 for everyone, Level 3 for a designated AI Lead in each school. An AI Growth Test twice a year for Grades 4–12.
3 Jun 2026
Every eligible private school gets one of two visits: a full inspection with a published rating, or a shorter monitoring visit that follows lines of enquiry from KHDA's data. Both come with no more than 24 hours' notice. KHDA picks the visit type from school data, including your Self-Evaluation Form.
From 2024/25
14 on governance and operations, 11 on teaching and learning, 14 on health, safety and well-being. All in force.
Recognise any of these?
Said the way school leaders say it, with where the evidence comes from.
“The SEF gets rebuilt before every inspection.”
KHDA visits now come with no more than 24 hours' notice, and KHDA reviews SEFs between visits.
RegulatorLives in Word, OneDrive
Pilot Live SEF
“Our SDP tracks what we did, not what changed.”
A pattern across four anonymised GCC inspection reports we've reviewed.
Client evidenceLives in Word, Teams
Pilot Outcome-tracking SDP
“Observation forms don't match the framework.”
The same four-school pattern.
Client evidenceLives in Word
Pilot Live SEF
“Middle leaders are drowning.”
TALIS 2024: admin and marking hurt well-being more than teaching. UAE teachers mark 5.4 hours a week against an OECD average of 4.6.
IndependentLives in Outlook, Word
Pilot Everyday drafting, Leadership continuity
“When people leave, the reasoning leaves.”
Two large UAE groups report staff attrition of about 9% and 17% (Khaleej Times, Aug 2026).
Company-statedLives in Teams, OneDrive
Pilot Leadership continuity
“39 ADEK policies, kept current by hand.”
ADEK's updated policy set, in force from 2024/25.
RegulatorLives in Word
Pilot Policy register
“We have to show ADEK responsible AI use.”
ADEK's AI policy: six safeguards and an AI Lead in every school.
RegulatorLives in Word, Teams
Pilot ADEK AI evidence file
“Staff use ChatGPT, but nothing reaches our documents.”
Gallup: only 19% of teachers work under a school AI policy.
IndependentLives in Outside Microsoft 365
Pilot Everyday drafting, ADEK AI evidence file
“Assessment data takes weeks to format.”
One GCC group's finance lead told us this.
Client evidenceLives in Excel
Pilot In progress
The pilot menu
Each pilot runs inside apps your staff already use, and each one has a number we measure before and after.
M1 + B1
Drafts your SEF with a source on every judgement, checks it against the framework descriptors, and keeps one live version.
B3 + M2
Decisions made in Teams become pending edits across the SDP, the governors' report and the staff brief. Activity actions get rewritten as outcomes.
J1 + J2
An AI use register (metadata only) and a self-assessment against ADEK's six safeguards and PDPL principles.
M3 + C2
Parent letters, meeting summaries and report comments in your house style, in Arabic and English.
M4 + F5
A handover file, the history behind key decisions, and a starter kit for new heads of department.
B4 + M5
Checks your 39 policies against the current ADEK versions and flags what has drifted.
J6
A two-page termly note, in Arabic and English, on what AI does in the school and what it never does.
Phase 2Coming once Excel and PowerPoint support ships: MAP export → RAG tracker and intervention lists, the termly board pack, and an AI impact scoreboard (a Word-only version is available now).
Not on the menu, by design: IEPs and anything near safeguarding.
Guardrails
Never
Always
How we line up with the safeguards ADEK published in September 2026. This is our mapping, not an ADEK endorsement. Use it as a checklist for any AI vendor, including us.
Proof
Case notes, not case studies: what was done, how it was measured, and where it stands today.
20+ pilots taken to production · 12B+ tokens processed
The offer
Every step is useful even if you stop there.
How much are you lagging behind AI leaders? Free
Three minutes. Your result shows on the page.
Pilot shortlist call Free
30 minutes. Your two best-fit pilots are already picked from the audit.
72-hour proof Free
Send us five documents. Within three working days we show one section redrafted in Word, citing your own evidence.
What we need: one SEF section, your current SDP, the policy list, a page of last term's report comments and the inspection framework section that applies. Microsoft 365 schools only, and a limited number each month.
Prove
One school, two or three workflows, one term. A written go/no-go at the end.
Prove + pre-committed scale
The scale-up price and scope are agreed up front, and the go/no-go triggers them.
Group partnership
All schools, shared rule packs, a team on the ground. A per-staff platform fee plus a fixed fee per workflow, with the projected token bill shown before go-live.
Prove · the paid pilot
Prove it in one school, in one term. Two or three workflows. A governance review before we touch any data. Baselines in three weeks, from documents you already produce. A written go/no-go against a bar we agree up front. If it doesn't clear the bar, you stop, and your data comes back under terms signed on day one. If it does, the pilot fee is credited against scale-up.
Pricing. Fixed fee, scoped in writing after a 30-minute call. Paid pilots are credited against later phases.
Keep it for everyday help. Copilot is a general assistant for each person ($18 per user a month on academic licences, with Copilot Chat included in A1, A3 and A5). We run in the same apps and add what Copilot doesn't package for schools: regulator rule packs as one-click checks, updates that flow into the documents that depend on a decision, a numbered source on every claim, edits only as tracked changes, shared memory across the school, and a team that sets the workflows up. Use Copilot for everyday help. Use us for the documents an inspector reads.
That's the demand, and it's a good sign. The problem is that chat work never reaches your controlled documents and leaves no register, which ADEK's oversight safeguard expects. If your school runs on Google Workspace: Gemini for Education is free for you, and our Google support is coming. Join the waitlist and start with the ADEK AI evidence file in Word.
We deploy in your own Microsoft 365 tenant. Workspaces keep HR and safeguarding data out. We work to the UAE PDPL (Federal Decree-Law 45/2021), ADEK's third-party vetting rules and, for Saudi schools, the KSA PDPL. Before any data moves, the governance review walks your IT lead and DPO through the data flows and every sub-processor. Your data is never used to train a model.
Out of scope, by design. We don't touch safeguarding case files.
Every claim cites a source you can open. Edits arrive only as tracked changes, a named person accepts each one, and rule checks flag judgements that have no evidence behind them.
No. It takes on the documents around teaching, not the teaching.
It drafts in Arabic and English, and your staff approve every draft. We test quality on your own documents during the pilot before claiming anything more.
There's no new app. It runs in Word, Outlook and Teams. ADEK's Level 1–3 training can use real work from the pilot.
That you have a register, a log and human sign-off, which is what the safeguards ask for. We never claim regulator endorsement, and you shouldn't either.
We work from exports today. A direct integration is custom work, scoped separately.
A fixed pilot fee. Before go-live, we show you the projected monthly token bill.
You get the founder's record, documents that stay as Word files in your own tenant, a deployment you can reverse, and exit terms in the contract.
Take the audit for a score and two recommended pilots. Or skip ahead and book the call.