Abuse protection players will actually install.
An app on the player's device. Its AI reads DMs, texts and comments on their posts right there, takes explicit abuse out of their view and names what is building.
The club, the league and Guardii never see the messages. The player decides what goes anywhere.
backedIt arrives by DM, and almost nobody reports it.
Betting has turned every result into a reason to message a player. Most of it lands in direct messages, where no club moderator can see it and few players will report it.
of NCAA Division I men's basketball players received threatening messages from someone who bet on their game.
NCAA, Feb 2026
of AFL players have been abused online. Only 4% reported it.
AFLPA Insights, Ed. 5
of NRL players targeted online did not report it.
RLPA Players' Pulse 2025
of direct abuse aimed at tennis players came from gamblers.
WTA & ITF
Players told us. We rebuilt it.
Our first version ran on servers, and players would not finish signing up. Nobody wants their club reading their DMs, however good the reason. So we rebuilt it: the AI runs on the player's device, and nobody, not the club, the league or us, can read their messages unless the player sends them.
The phone spots the pattern.The player makes the call.
Everything happens on the player's device. Nothing reaches welfare, integrity or police unless the player sends it, along routes they agreed at sign-up.
The player opts in
Account by account
The player, and family if they choose, installs it and connects the accounts they pick. Any account can stay off.
Checked on the phone
Only the apps they connect
DMs, WhatsApp, texts and comments on their posts are read by AI on the phone itself.
Removed from view
Named as it builds
Explicit abuse is taken out of the player's view. Approaches that build over time, like betting pressure, are named.
The player decides
Block, hide or send
Block the sender, hide it, mark it harmless, or send a sealed evidence package to the people they chose at sign-up.
Welfare, if chosen
People make the call
Welfare sees how serious it is and how sure the software is, through routes the player agreed. Distress goes to care, never to an investigation.
Their phone, their accounts, their call.
Comments on their posts
Abuse comes off the feed, and is named and explained in the player's inbox so nothing disappears without a trace.
DMs, WhatsApp and texts
Checked on the phone. Threats, betting harassment and approaches about inside information are named for what they are.
Evidence when they want it
A tamper-proof evidence file, sent where the player chooses: security, integrity, a clinician or police.
Their settings
Accounts on or off, who is told, and what counts as abuse for them. Rivalry banter stays banter.
- A record of protection, not of messages: how many players are protected, alerts raised and escalations.
- Reports a player chooses to send to welfare, with severity and confidence, never the words.
- Routes for threats and coordinated pile-ons, agreed with the players' association before anyone signs up.
- No message content held anywhere, so nothing for the club to be asked for later.
What the player chooses to send.
Guardii runs on the player's device. When a player reports, welfare sees how serious it is and how sure Guardii is, with the evidence they chose to share.
AssistantSent by players onlyTwo players chose to send something to welfare this week. You see how serious it is and how sure the software is, never their messages.
PlayerSet up with your team. Reports land in your systems.
Our engineers work alongside your club, league or players’ association to set who hears what, and when, around how you already work. Reports arrive in the tools your people already use.
Your AI agents
Reports arrive through a secure connection your AI tools can query with your credentials, so they can help with the casework.
Your systems
Or straight into the case management, HR and reporting systems you already run, so a report lands where your people already work.
You see only the reports people send. There is no search across everyone, and every time a report is opened, it is logged.
Your roles become the access rules
Who is told what, at what threshold, and who is never told, set with you before anything goes live.
Your policy, versioned
Who hears what, and when, follows your policy and your local law. Every change is recorded and can be rolled back.
In your region
Deployed in the region whose rules apply to you, with records held there.
Set up with you
A working session with your team, then a small pilot. Our engineers work alongside yours while it is set up.
English and Arabic, dialects included.Checked by native speakers.
Guardii works in English and Arabic today. Every language we add is checked by native speakers before it goes live, which is the part that takes time.
English
Slang, abbreviations and emoji included.
Arabic
Modern Standard and Gulf dialects.
Mixed messages
English and Arabic in the same chat, and Arabic typed in Latin letters.
Private by design.
Uptake depends on trust. Guardii does not monitor the people it protects.
Messages stay on the device
Detection runs on the device. Guardii never receives messages, so there is nothing to access or browse.
Opt-in and disclosed
No one is covered without knowing. Each person chooses which accounts are covered and who is alerted.
Alerts, not transcripts
Only an alert, or evidence the person chooses to send, leaves the device. Never a message feed.
Bring your players' association.
We work through consent with the people who represent players, from the first conversation.