Case study
Workplace AI safety tool
A browser guard that checks what employees paste into AI tools and blocks private data before it leaves the company.
Data-loss protection for staff using AI
- Before
- the data leaves, not after
- Tax IDs
- and other private numbers flagged
- 0
- false alarms on the strict test set
The problem
Staff already use AI tools at work. Most companies cannot see what is being pasted into them.
What we built
- A browser extension that screens prompts before they are sent.
- A team dashboard showing which AI tools are used, without storing what people wrote.
- An output filter and red-team test suite for AI apps the company builds itself.
Privacy
The dashboard keeps counts and categories, never the text of a prompt.
What it taught us
Staff will use AI either way. The job is to make the safe way the easiest one.