Autonomous safeguardingfor every student communication.
Detection-to-pastoral-action, autonomously. Guardii processes student communications, classifies threats — bullying, grooming, exclusion, sextortion, self-harm — and routes only what your safeguarding lead needs to act on. Humans are engaged only when their judgment is decisive.
backedThe volume has outpaced the safeguarding workforce.
Cyberbullying, exclusion campaigns, targeted harassment, and predatory grooming now happen across Instagram, TikTok, Snapchat, Discord, and WhatsApp — platforms schools have no visibility into. The work that used to require a safeguarding workforce is now done by software. Humans are called only when the call matters.
Schools with diverse, multicultural student populations face additional complexity — bullying that crosses cultural and language barriers, nuanced social dynamics across dozens of nationalities, and behaviour patterns simple keyword filters cannot detect. Guardii operates across every language a student speaks.
Students experience cyberbullying
Of bullied students never report
Rise in online grooming since 2020
Of incidents start on social media
The system does the safeguarding.The pastoral lead is engaged only when judgment is decisive.
Detection, classification, and pastoral packaging run autonomously. Your safeguarding lead is engaged only at the decision point — where their judgment matters most.
Ingestion
Student communications
DMs, group chats, and public comments ingested across student social media. No human review.
Detection
Behavioural ontology
Bullying, exclusion campaigns, grooming, and self-harm signals scored in context. No human review.
Classification
Severity tier
Each signal tiered by severity, urgency, and pastoral fit. Routed automatically. No human review.
Pastoral Packaging
Context, not transcripts
Safeguarding context compiled — what the lead needs to act, not every message. No human review.
Lead Notification
Human in the loop
Routed to your designated safeguarding lead. Human judgment is engaged only at the decision point.
Safeguarding-as-a-service. Autonomous by default. Pastoral-lead-in-the-loop by design.
School Safeguarding Dashboard
District-wide visibility across every student — duty of care, automated.
Student Protection Status
Last 7 daysLive Alert Feed
Grooming pattern detected — Year 8 student, adult contact escalating across DMs
Cyberbullying escalation — coordinated exclusion campaign across 3 accounts targeting Year 7 student
Sextortion attempt blocked — evidence preserved and escalated to safeguarding lead
Repeated targeted harassment flagged — racial abuse in group chat messages
Wellbeing check — student sentiment shift detected, counsellor notified
Anti-Bullying Detection
Detects cyberbullying, exclusion campaigns, targeted harassment, and coordinated abuse — across comments, messages, and group chats in real time.
Automated Safeguarding Escalation
Alerts route directly to your designated safeguarding lead, school counsellor, or district safeguarding team — no manual monitoring required.
Duty-of-Care Compliance
Automated weekly and term reports for school leadership, governing boards, and district administrators on safeguarding trends and actions taken.
Behavioural intelligence — not a keyword filter.
The system detects and classifies harmful behaviour autonomously. A proprietary behavioural ontology understands intent and context — not keywords. Subtle exclusion campaigns, coded language, and trust-building sequences are surfaced without human review at every step.
Cyberbullying — harassment, name-calling, threats, and targeted abuse across comments, messages, and group chats
Exclusion campaigns — coordinated social isolation, rumour spreading, and pile-on behaviour
Predatory grooming — trust-building, isolation tactics, and platform migration attempts by adults
Sextortion and coercion — image-based threats, blackmail, and exploitation attempts
Self-harm signals — concerning language patterns indicating a student may be at risk
Identity-based bullying — racial, religious, coded language, and cultural nuance across 40+ languages
Routed direct to the safeguarding lead.
The system performs detection, classification, escalation, and pastoral packaging autonomously. Verified concerns route directly to the designated safeguarding lead — with the right level of detail. Parental involvement is configurable; the school controls the trigger.
Safeguarding Lead Escalation
Critical concerns route to your designated safeguarding lead with full context. Configurable to your policy.
Privacy-First Monitoring
Surface what the lead needs to act — not every message a student sends. Pastoral, not surveillance.
Multicultural & Multilingual
Detects bullying and grooming patterns across 40+ languages, dialects, and cultural contexts.
Parent Involvement Controls
Schools control when and how parents are notified. Community-based protection without unnecessary alarm.
Duty-of-Care Reporting
Automated compliance reports for governors, district leadership, and inspectors — evidence-grade.
Preventive, Not Reactive
Foresight into behavioural trends — at-risk students surfaced before incidents require police involvement.
Where students communicate, Guardii is deployed.
DMs, group chats, and public comments across the platforms students actually use. Native integrations and roadmap expansion across the full social-media surface.
Every language a student speaks.No locale out of scope.
Critical for international and multicultural schools. Detection models are fine-tuned across 40+ languages, regional dialects, and code-switching — with cultural context for bullying patterns that don't translate.
Detection // Classification // Escalation — in any language the population speaks.
Safety and surveillance are not the same thing
One watches people. The other watches for harm. Most tools sold into this category blur that line, and every time one does, it makes the people it was meant to protect less safe, because they route around it. Guardii is built so the distinction holds under pressure.
Patterns, not transcripts
Detection reports the shape of a conversation — sustained targeting, escalation, coercion — not its contents. You are told what is happening and how serious it is. You are not handed a feed of what was said.
Nobody reads it, including us
Content is committed in a form we cannot browse or mine. We hold the integrity of the record rather than access to it, which is what lets the same system protect a person and preserve evidence without reading either.
The person protected knows it is there
Guardii is not installed behind someone’s back. A child, a student or a worker is told what is covered. Protection people discover rather than consent to is the thing that destroys trust, and trust is what makes them keep talking.
The subject is the client
Alerts serve the person at risk and the adult responsible for them. Where an organisation pays, it receives what it needs to run the programme, never a window into somebody’s private life.
Least data, shortest path
We take the minimum required to detect the harm, and we do not repurpose it. No profiling for advertising, no resale, no secondary use dressed up as product improvement.
This is a design constraint rather than a policy preference. A system that cannot read what it protects cannot be quietly repurposed into one that does, by us or by anyone who acquires us.
Engage the deployment team
Ready to evaluate a pilot for your school or district? Reach out directly.
// Direct
cam@guardii.ai