Product

Decision support your counsellors can defend

MindGuard helps trained practitioners identify early signs of distress in consented digital content — powered by Mental-RoBERTa, reviewed by humans, built for trust.

How MindGuard works

A 4-stage consent and decision-support pipeline designed for school and university counsellors.

Set up your roster

Upload your student roster once. MindGuard determines adult vs. minor status and routes consent requests to the right person — the student, the parent, or both.

Consent is requested and recorded

Students and parents receive a signed, one-time consent link. Every view, accept, decline and revoke is timestamped in an immutable audit trail.

Monitoring runs only for consented students

For consented students, MindGuard analyses content they explicitly share and produces risk summaries a trained counsellor reviews — never an automated decision.

Counsellors act, with crisis resources

High-risk signals surface in a rolling risk view with built-in crisis resources by country and US state. The counsellor remains the human in the loop.

One wellbeing view across digital touchpoints

Preview how MindGuard securely processes signals across Google Docs, Canvas LMS, School Email, and consented social channels.

One wellbeing view

Across the digital spaces students use

Illustrative example — no real student data. Every connector is opt-in and consent-gated.

Public forum post (illustrative)

OAuth opt-inOAuth verified

"Late night thoughts... feeling completely isolated and losing hope..."

Signal

High distress flag

Review queue

Summary → Counsellor

Counsellor action

Formatted summary to triage

Safeguard note: Requires an explicit OAuth token handshake from the student.Open the app →

Consent-first philosophy

Consent is the product

We built the line we are willing to stand behind. MindGuard is not a monitoring wiretap — it only analyses content a student explicitly chooses to share, and only after consent is in place. AI detects signals. Humans provide care.

Clear, revocable consent is recorded before any analysis
Adults consent for themselves; parents consent for minors
Withdrawal is one click and stops analysis immediately
The human counsellor reviews every output

Human oversight

Every output reviewed by a counsellor.

Explainable

Summaries a professional can defend.

Early signals

Meaningful patterns, surfaced sooner.

Built for care

Starting points for real conversations.

Our commitment — in four parts

Everything MindGuard does is governed by four safeguards we will not compromise.

Consent-first

No content is analysed until consent is given — by the student if adult, by a parent or guardian if minor. Consent can be withdrawn with one click, and withdrawal stops analysis immediately.

Human-in-the-loop

MindGuard is decision support, not a decision maker. It surfaces signals and suggests crisis resources; a trained counsellor reviews every output before any action.

FERPA / COPPA aware

Built for the regulatory reality of schools: student records are protected, minors are routed through parental consent, and institutions stay in control of their data.

Data minimisation

Only content a student explicitly shares is analysed. PII is encrypted at rest, the analysed content is not stored between sessions, and data is never sold or shared outside the school.

What MindGuard is

An early-signal tool that helps counsellors prioritise follow-ups
A consent-first workflow with a full audit trail
A structured, evidence-based starting point for conversations
A 4-tier risk view (low / moderate / high / critical) with crisis resources

What MindGuard is not

Not a diagnosis tool — it never labels a student
Not a monitoring wiretap — only explicitly shared content is analysed
Not an automated decision maker — every output needs human review
Not a replacement for counsellors, crisis services, or clinical care

Under the hood

A purpose-trained model, reviewed by humans.

Mental-RoBERTa Model

A transformer pre-trained on millions of mental-health domain posts and fine-tuned on 12,656 annotated examples. ROC-AUC 0.9813, 92.5% accuracy.

Human-reviewed outputs

Risk scores surface as summaries for counsellors — the human keeps the decision. No automated escalation, no un-reviewed flags.

Session-only analysis

Analysed content is not stored between sessions. Roster PII is encrypted at rest and access is limited to institution-authorised staff.

See MindGuard in action

Request a dedicated institutional walkthrough with your data, or launch the app directly to explore.

Request a demo Launch App