Letterlock The most secure AI assistant for Gmail

Responding to emails is a pain. Many tools are available, but when you choose an AI assistant entrusted with your email, you are risking the privacy of you and everyone that emails you. Letterlock solves both problems at once: Letterlock waits for emails and securely searches your inbox and Google calendar for relevant information before drafting a reply, saving hours a week in drafting replies to emails.

» Connect your Gmail    How it works →

How secure is Letterlock?

With ever more powerful AI agents, email security requires several layers of defense. Letterlock uses the same security protocols as Signal messenger and Swiss Banks, to ensure defense-in-depth. Email data is never saved on Letterlock servers. Compute is done in secure enclaves, and Letterlock's security algorithms are constantly monitoring for suspicious activity. Email access tokens requires simultaneous access to an encrypted secure enclave as well as Letterlock's EU-based verifier server. As a separate layer of defense, personally identifying information or private information is stripped before AI inference is run on a separate, secure enclave inference provider, which verifiably cannot store your email data.

Why doesn't Letterlock send emails?

The world doesn't need more AI slop. Letterlock never sends email, and never will. Instead, Letterlock does 80% of the work so you can do the remaining 20%. The Letterlock assistant securely gathers relevant information from your email and provides a draft email, lets you know when you missed an important email, makes sure all your appointments get on your calendar, and fixes grammar or spelling while fact-checking your email drafts as soon as you type @llfix in your email draft. Lastly, Letterlock matches the length of your emails, and is configured to avoid overclaiming or hallucinating email content. If Letterlock doesn't know something, it won't write it.

Features

A Trusted Execution Environment is a locked compartment inside a processor chip. Code or data inside cannot be seen or modified by the host operating system, the cloud provider. At boot, the enclave produces a signed attestation report containing a hash of the code it is running. Anyone can verify that hash against the published source code. Letterlock goes a step beyond: Even within the TEE, all data are encrypted using a separate signing server. Even processes within the enclave cannot see enclave data without permission from a separate, monitored signing server.

See pricing →     Read the FAQ →