Data stays on your device
Manuscript state, marks, dictionary entries, scan state, settings, and your API key all live in your browser's localStorage. We do not store any of it on our servers.
Privacy
Your manuscript stays yours.
KRW has no accounts, no server-side database, and no analytics or tracking scripts. Everything lives in your browser and nowhere else.
When text leaves your browser
Only when you scan, interpret, or full-scan. The relevant text plus your dictionary context is sent securely to the AI provider you selected.
Your API key
Stored in localStorage alongside your other settings. Sent with each request to the AI provider over HTTPS. Not stored server-side, and server logging is disabled.
Your choices
Clear localStorage to delete everything. Export your data from the gear menu. Remove your API key anytime. Use demo mode without connecting a provider.
How the data path works
Three layers. You control the first two.
Your browser
All data is created and stored locally. Nothing leaves unless you trigger a scan or interpretation.
KRW's server
Stateless. No logging. Your request is forwarded to the AI provider and the response is returned. Nothing is stored or persisted.
Your AI provider
Governed by the provider you choose. We link to their policies below so you can read the terms yourself.
Provider policies
What happens at the AI layer.
Provider
Google Gemini (free tier default)
Retention
55 days for abuse monitoring
Training
Free tier: may use content for improvement. Billing-enabled: opt-in only.
Policy
Google's usage policies
Provider
Anthropic Claude
Retention
Up to 30 days, then deleted
Training
Never used for model training
Policy
Anthropic's data policy
Start reading
No account needed. No data collected. Just open and read.
KRW controls the browser and server layers. The third is governed by the provider you choose.