Privacy & Your Data
Premail has no servers processing your private data. There is no Premail cloud, no Premail account, and no backend that your email passes through. Everything runs on your computer.
The one exception is payment: if you upgrade to Pro, checkout is handled by Stripe. Stripe requires your email address to process the payment and send a receipt. That email address goes to Stripe, not to us, and is never tied to your inbox content or activity.
Do your emails ever reach Premail?
When Premail monitors your inbox, it connects directly from your machine to your email provider (Google, Microsoft, or Apple). Nothing goes through any Premail infrastructure, because there isn't any. The people who built Premail cannot see your email, your rules, your contacts, or anything else. Technically impossible, not just policy.
What does leave your machine?
To classify a message, Premail sends the sender address, subject line, and message body to whichever AI provider you've configured. If you chose Anthropic or OpenAI, that content travels to their servers under their respective privacy policies. Your API key goes with that request but is never sent anywhere else.
Anonymous product analytics is off by default. You can choose to enable it under Your Plan. If enabled, Premail sends bounded feature milestones and counts to PostHog. It never sends email content, addresses, account IDs, API keys, rule text, custom domains, or license tokens. Turning it off stops capture and deletes the local anonymous analytics identifier and transmitted milestone markers. A local first-open timestamp remains on your device so a later opt-in can measure retention from Premail's measurement baseline; it is never transmitted while analytics is disabled. Premail does not send crash reports.
What happens if you use Ollama?
If you configure Premail to use Ollama, the AI model runs locally on your computer. Email content stays entirely on your machine. Nothing is sent to any external server for classification. Not to Premail, not to Anthropic, not to anyone.
This is the fully air-gapped option. The tradeoff is speed and accuracy: local models are slower and vary in quality, but for people who need their email to stay private end to end, it's the right choice. We recommend the qwen3 model for best results. See the AI Model Configuration article for setup steps.
Still have questions?
Contact Support (for Pro members)