Challenge a charge.
Share the charge. Leave out your account number.
ON YOUR SIDE.
Too much
information.
Give AI what it needs.
And no more.
Prepare sensitive text before sharing it with AI. Keep the useful context. Leave out personal details.
No account needed. Text processing stays in your browser.
FAMILIAR TASK. EXTRA INFORMATION.
More context.
Fewer personal details.
Improve a proposal.
Share the scope. Replace client identities.
Find the right words.
Share the situation. Leave out names.
A SMALL STEP BEFORE YOU SHARE.
Same AI. Less oversharing.
IN YOUR BROWSER
Your original
Maya’s £45 bill
/
Prepare & review
[PERSON]’s £45 bill
YOUR CHOSEN AI
Get help
Copy only the reviewed text.
BACK IN YOUR BROWSER
Restore locally
Bring the answer back.
Put your details back in.
Illustration · you copy between PRE/LLM and your AI. Nothing is sent automatically.
YOUR DATA. YOUR SAY.
Your original stays
on this device.
Text checks happen in your browser. No text is uploaded to PRE/LLM servers or cloud AI providers.
You review. You choose what to share.
Try a fictional example ↗What works today—and what doesn’t?
Today: local pattern checks, not an AI model. Pasted text, basic email/phone detection, exact details you specify and local restoration. Names aren’t detected automatically. Documents and voice aren’t supported yet.
The / on this website is a demonstration, not active protection. The browser extension remains experimental. PRE/LLM does not currently control access or actions across your AI apps.
Masking isn’t guaranteed anonymity. Context, browser extensions, keyboards, clipboard history and device security can affect privacy. Temporary mappings expire after 15 minutes or disappear when you close or refresh the tool.