Apps on Quad can be built by anyone, anywhere, but real school data lives in one place only: your school's own environment, in your region. Builders work on a fictional school. Quad reviews code, never data. And nothing touches a real record until your school says yes.
Quad keeps building and real use apart by design. Where an app is built never holds school data, and where it runs is your school's alone.
When an app moves to a school, only its code moves. Sharing, installs and copies never carry data, credentials or approvals with them.
Your school's region is chosen when it joins Quad, to meet the data residency rules that apply to you. Your school's data stays in that region.
Quad works for schools anywhere, so the public platform where builders work is hosted in the United States. It never holds school data.
| What | Where it is held | Who can see it |
|---|---|---|
| Your school's dataStudents, staff, parents, classes and groups, and timetable, register or medical information where your systems supply it | Your school's environment, in your region | Only apps your school has approved, for the data they were approved for, and only what each signed-in person is allowed to see |
| What apps storeThe records an app keeps, such as consent forms or reading levels | Your school's environment, in a separate table for each app | That app alone, under its own access rules |
| AI building by your staffEnterprise only | Claude on Amazon Bedrock, in your region's geography. For Australian schools, within Australia | It works on app code and simulated data. It never reads real school data |
| What Quad reviewsFor each app release | Quad's review console | Quad's reviewers see code, the app's manifest, screenshots from the fictional school and the security scan's report. Never school data |
| Builder accounts and app codeOn the public platform | AWS, United States (us-east-1) | The builder and the people they invite to a project |
Regions open as schools join, once Claude's availability there has been checked. For example:
Quad is built in layers. We run on independently audited infrastructure, and add our own isolation on top, so that safety never depends on an app, or the AI that wrote it, behaving well.
Quad runs on Amazon Web Services. Each region's school environment has an AWS account of its own, apart from the public platform, and each Enterprise school's environment has its own account again. AWS's services and data centres are independently audited, including to SOC 2 Type II and ISO 27001.*
Schools in a regional environment share code and infrastructure, never data. Every record carries its school's tenant id and is scoped to it by default, and school data is protected by row-level security in the database. Each school has its own secrets for its sign-in and system connections, its own rate limits, and its own export and deletion.
Every app at every school has its own data table, locked by an access policy on that app's own functions, so one app can't read another's data, even through a bug. Apps have no server code of their own, and never receive database credentials, cloud credentials or anyone's session. Scheduled jobs run in an isolated function with no route to the internet.
Apps run on their own domain, quadapps.net, never on Quad's, each at its own address and inside a sandboxed frame. Their content security policy lets them load code and send data only to Quad's API and package CDN, so an app can't send your data anywhere else. The domain is HTTPS only, with HSTS preload.
A release is checked and reviewed once, and every school then takes that same checked release. Nothing touches real data before your school approves it.
Every package an app adds, and everything it depends on, is pinned and checked: no install scripts, a malware and known-vulnerability scan, and a licence check. Packages are re-scanned when new advisories appear.
Policy checks catch banned modules, raw queries, secrets written into code and data sent outside the SDK. Tests run, and the release is built in an isolated job with no network. An AI security scan adds findings for the reviewer.
A Quad reviewer looks at the code, the app's manifest, screenshots from the fictional school and the scan's report. Reviewers never need, or get, access to your school's data.
Your IT lead sees one screen, in plain language: what the app reads from your systems, what it stores, and what it writes back. Only a school admin, signed in with your school's SSO, can approve it.
Each school runs a pinned release and chooses when to upgrade. If an upgrade changes what the app reads, stores or writes back, or who can use it and see what, your school approves it again.
If a release or package is found to be vulnerable, Quad can suspend it at every school running it, and tell those schools.
Each release comes with an assurance report: what it reads, stores and writes back, how it's isolated and where it runs, and the checks it passed.
Beyond the infrastructure, the platform itself enforces who sees what. Apps get their data only through Quad, so these rules apply to every app, whoever built it.
People use apps with your school's own accounts, through single sign-on. There are no separate passwords to manage, and apps are internal to your school: there is no guest access.
Every field has a sensitivity level, from basic details to contact, academic, pastoral and medical, and an app asks for exactly the levels it needs. A reading tracker can't see medical notes because it was never granted them. Staff see all students, a parent sees only their own children, and a student sees only themselves.
Where your school's system records that a parent can't be contacted about a child, for example under a court order, Quad applies it to every notification and task from every app. Apps can't opt out, and every suppressed message is audited. Where your system doesn't record restrictions, every linked parent is treated as contactable, and Quad tells you so when you connect an app.
Every read of pastoral or medical information is recorded, along with tasks, notifications and scheduled job runs. The record can't be edited, and your admins see it in your school's security centre.
Data is encrypted in transit and where it is stored. School databases keep point-in-time recovery, and an app's data is backed up before an upgrade changes it, so a school can roll back to the previous release.
AI helps build apps; running one never uses AI. Quad's AI builder works only on app code and simulated data, and app data is never sent to it. Model access is held by Quad's AI gateway, never by the build environment, and under Amazon's terms for Bedrock your requests aren't used to train models.
Your school's data belongs to your school. Quad uses it only to run the apps your school has approved. We never sell or monetise school or student data, use it for advertising, or profile anyone.
A standard data processing agreement comes with setting up your school's console, and a formal signed version is available on request.
Your admins stay in control: they can suspend any app instantly, revoke what it can see, export its data, and see who has opened sensitive information.
We're happy to answer any question about how Quad handles your school's data. If you're an IT lead or privacy officer reviewing an app, start with its assurance report, then email us for anything it doesn't cover.
* Certifications mentioned on this page are held by Amazon Web Services, Quad's cloud infrastructure provider, not by Quad itself.
Quad handles personal information in line with the Australian Privacy Act 1988 and the Australian Privacy Principles. How we handle builders' own account information is set out in our privacy policy.
It starts free, it takes an afternoon, and nothing goes near your school's data until you say so.