Guardii protects your people from psychosocial harm online.
It is an AI safety platform that connects to the chat and messaging apps they already use. You choose what it watches for.
The whole thing in four steps
It goes on the person's device, where they connect the chat or social apps they need protected, privately.
Messages are classified as they arrive. The AI runs on the device, so no one, including Guardii, the employer or any third party AI service, can see the messages.
Explicit abuse is flagged instantly. Escalating harassment triggers a warning.
The employee decides what happens next, and who receives the evidence report, if anyone.
Thresholds are set by the employer and the individual, along with the escalation pathways: who is told, and at what point.
See it workStep through one conversation, from the first message to getting support Open ›
Two kinds of harm, handled two ways
An explicit image, a threat, a racial slur. It is overt, so Guardii flags it in the Guardii app the moment it arrives.
The person sees what it is and chooses whether to report it.
A manager messages a junior employee late at night. First compliments, then an offer to help with a promotion, then pressure to meet alone and keep it quiet. No single message is abusive. Together they are harassment.
Guardii reads the whole conversation over time. Once it starts to look like harassment, the person gets a warning that says what it looks like. If it gets serious, and only if they agreed at sign up, the role you nominated hears too.
Everything else is untouched and fully private. The messages stay on the person's device. Nothing leaves it unless they choose to file a report. That is why people install it, and uptake is what makes every other number real.
You choose what it watches for
Turn on what matters to your organisation, leave the rest off.
The same product, different organisations
| Organisation | What they turn on | Who hears about it |
|---|---|---|
| A mining workforce | Sexual harassment, bullying, coercion on remote sites | People and culture |
| A professional firm | Harassment of staff, coercion, bribery | Risk and conduct |
| A football club | Threats and racial abuse in comments and messages | Player welfare |
| A cricket body | Match fixing approaches, inside information | Integrity unit |
| Their families, as a benefit | Grooming, sextortion and bullying aimed at the kids | The parent |
Escalation pathways: distress and signs of self harm go to a counsellor, a doctor or a mental health service. Harassment and bullying go to people and culture or the wellbeing team. Exploitation of a child goes to child protection. Only a crime goes to police, and only then is a sealed evidence package built.
Nothing leaves the device unless the person agreed to it at sign up, at a threshold the organisation set and the person could see.
Why it matters to the claim
Australian workers sexually harassed at work in the last five years. Half of it repeated, half of that for over a year.
of it reported. Most of it is never seen by the organisation at all.
median psychological injury claim, 35.7 weeks off work. Claims up 161% in a decade.
Australian Human Rights Commission, Time for Respect, 2022. Safe Work Australia, 2025.
Good to know
Completely private. Nobody can read your people's messages, including the employer and us. A nominated person is told a concern exists; what was said stays with the person.
Reads the whole conversation. Over days and weeks, the way a person would notice it.
Works with what you have. It picks harm up as it arrives and can feed the reporting and case management tools you already use.
Across their apps. Texts, messaging apps, email, direct messages and the comments on their posts, on any device.