UNVEIL

A workspace where people and AI agents work together. A person stays in charge, and a guard that cannot be switched off checks every action before it runs.

Scroll to follow one action through the checks
Start · the app

An agent wants to do something

Read a file. Open a web page. Run a command. In UNVEIL every one of these starts as a request. Nothing runs yet.

Check 1 · the request

Does it match what you asked for?

The app compares the action with the task you gave. An agent that reaches for something outside the task is refused here.

Check 2 · you

You say yes or no

Actions that touch your files, your accounts or the outside world wait for your approval. You can allow once, allow for the session, or deny.

Check 3 · UNVEIL guard

The guard checks last, even after a yes

Before anything runs, the guard looks at the action itself for signs of misuse or malware. If it finds them, the action is blocked, whatever was approved before it.

The guard can only say no. It has no off switch.

Always on · background watch

It keeps watching when you are away

The guard re-checks the workspace every 60 seconds, with or without anyone signed in, and every decision is written to a signed record.

The app

All of it sits inside one app

Works in a browser and installs to a phone home screen.

app.unveilresearch.com

Access is by invitation only. Public sign ups, coming soon.

Why we built it

Agents can act on their own now. The safety around them should not be optional.

AI agents read files, browse the web and run commands without a person typing each step. On most platforms the protection around that is a settings page, and settings get turned off.

UNVEIL puts the protection in the core. The aim is a place where people and agents work together, not against each other, with human judgment as the final authority.