Anthropic’s AI assistant Claude has gained a new ability: once a user asks it to reply to a message, the system can now handle everything from drafting the response to actually pressing the send button in Gmail. That means the assistant will not always ask for a second confirmation before sending, so using the feature can require a degree of trust from the user.
The Claude connector has previously allowed the AI to integrate with Google services such as Gmail, Calendar, and Drive, but sending emails from Gmail was not enabled until now. With this update, the assistant still requires the user to request that it reply, but after that instruction it independently composes the reply and sends the message without further prompts.
The setting is optional: users who prefer to review the AI’s suggested text can keep the manual review step, which remains the default behavior. According to 9to5Google, Anthropic is initially making the capability available in paid Claude tiers.
The change raises both convenience and security considerations. While automatic sending can save time for people who frequently ask the assistant to generate short replies, the lack of a mandatory final confirmation may concern some individuals and organizations. Anthropic’s optional review mode preserves user control, but real‑world adoption will depend on user preferences and institutional policies.
In short: Claude can now, upon user request, draft and automatically send Gmail replies; reviewing the message before sending is optional, and the feature is reportedly rolling out first to paid versions, per 9to5Google.
