All demosSend Guard
Send Guard
Check a message for exposed secrets, personal data, risky commitments and tone before sending it.
Powered by/yes-no/classify/rate
Performance & cost
This page session · since page load
- Typical request time
- —
- Input tokens
- 0
- Estimated cost · USD
- $0.00
How these numbers work · 0 successful requests
- Model compute
- —
- Successful requests
- 0
Average per measured successful request.
Run the demo to measure live requests.
Request time is the median time for a successful request, including queue, network, and retries. One request can judge many inputs: model compute adds their processing time and can exceed browser time.
Input tokens come from API response headers. Cost is estimated at $0.04 per million input tokens, before plan credits. Missing usage is marked unavailable or as a partial total (≥). Demo resets keep these totals; reload the page to start a new session.
Write a message or select Start replay, then inspect the Send verdict and highlighted text.
How it works
After a typing pause, the model checks the draft and relevant text fragments. Channel context and local policy combine the results into an allow, warning or block decision.
Demo scope
This is a simulated chat composer. Audience-fit scores are advisory and can miss a mismatch. The guard can make mistakes; a green verdict does not guarantee that a message is safe to share.
Inspired by the work of dabit3.