The safety layer between kids and AI.
GuardRail scans every message before it reaches a child. TrustEd shows their parent exactly what happened next — so parents are never the last to find out.
Real examples, run through GuardRail's real default rules on a loop. Click the box to try your own.
Result appears here as you type.
0M+
K-12 students in the US¹
0+
States actively legislating AI/child online safety²
<0h
Design target for GuardRail integration³
¹ National Center for Education Statistics, public + private K-12 enrollment. ² Based on public legislative tracking; changes as states introduce or pass bills. ³ A target for a single REST integration, not yet measured across live customers.
No dashboards to dig through
Ask Guardia a question. Get a real brief back.
Parents and district admins can ask in plain English — “Any concerning topics this week?” — and Guardia reads the real, already-scanned activity and answers in seconds. Every number in the brief comes straight from GuardRail's logs, never invented.
Ask Guardia
Real activity in, a plain-English brief out.
How it works
From integration to audit trail, in three steps.
Point traffic at GuardRail
One REST endpoint. Any AI app, any language. Built to integrate in under 4 hours.
Every message gets scanned
Content is scored against the grade band's policy in real time — blocked, flagged, or passed before the child sees a reply.
Parents see what happened
TrustEd turns every flagged moment into a full transcript a parent can actually read, react to, and act on.
One score, every platform
See a district's real-time risk posture at a glance.
Every scan rolls up into one weighted score and a live breakdown by severity — the same view a district admin sees inside GuardRail.
4
Risk tiers scored
3
Grade bands supported
scan
Updated per
9
Categories tracked
2
Low risk
GuardRail + TrustEd
Everything a district, a vendor, and a parent each actually need.
Real-time detection
Every message is scored before it reaches a child — not sampled afterward.
Age-tiered policies
K-5, 6-8, and 9-12 each get their own rules. Districts edit them live, no redeploy.
Audit-ready reporting
Every scan is logged and exportable, formatted for state child-safety statutes.
Full session transcripts
Parents see the whole conversation, not a single flagged line stripped of context.
One-tap consent
When an app wants a child's data, a parent approves or denies it from their phone.
Time & subject controls
Per-app daily limits parents set themselves — no ticket to the district required.
Who it's for
One platform, three very different needs.
Districts
Compliance that doesn't need a redeploy.
Set age-tiered policies once. Edit keyword rules and actions live from GuardRail — no vendor ticket, no code change.
- Audit-ready exports for state statutes
- One weighted risk score per district
- Microsoft Entra ID SSO support
Parents
The full conversation, not a flagged fragment.
TrustEd shows real session transcripts, screen time, and one-tap consent requests — from your phone, in plain English.
- Full transcripts, never a stripped line
- Ask Guardia a question in plain English
- One-tap approve or deny a data request
Ed-tech vendors
Ship compliance without building it.
Point your AI app at one REST endpoint. GuardRail scores every message before your model replies — no ML team required.
- One REST endpoint, any language
- Scores content before the reply is sent
- Built to integrate in under 4 hours
Security & privacy
Built with real protections, not a checkbox.
TLS in transit
All traffic between apps, dashboards, and servers is encrypted.
Encrypted DB connection
Our database connection requires SSL.
Rate-limited endpoints
Pairing and ingestion endpoints are throttled to reduce abuse.
Scoped access only
Only a student's linked parent and their district admins can view their data.
Full details in our privacy policy.
Questions
Before you ask, a few answers.
Gemini, ChatGPT, and Claude. On Android, a real Accessibility Service reads on-screen text from just those three apps. On iOS, Apple's platform doesn't allow any app to read message content — so the iOS companion tracks usage time only, never what was said.
No — one REST endpoint. Any AI application, any language, points traffic at GuardRail and gets a risk score back before the reply reaches a student.
Only the parent or guardian linked to that student, and authorized administrators at that student's district. Guardia doesn't sell data or use it for advertising.
No. Apple's Family Controls framework doesn't permit any app — including ours — to read on-screen content on iOS, at any price tier. The iOS companion only reports usage-time thresholds.
The reply never reaches the child. The attempt is logged as a scan event with its risk tier and matched category, and shows up in both the district's GuardRail alerts and the parent's TrustEd dashboard.
“Every child deserves an AI experience that is safe, transparent, and worthy of their parents' trust.”
Learn about Guardia →No dominant compliance vendor yet. That window closes fast.
Ohio's framework is already becoming the national template. Districts are looking for a solution today.