Claude can now work inside Gmail. Anthropic’s Gmail connector lets Claude search emails, read messages, draft replies, send and forward email, and organize an inbox with labels and threads. The connector is available on all plans, including free plans, which removes the usual paywall from the most practical part of the feature.
Connecting Claude to Gmail requires signing in with a Google account and granting permissions. Once connected, users can ask Claude to find emails, draft replies, or handle inbox tasks without learning special commands. Claude can also apply labels and arrange threads on its own, giving the assistant more control over the unglamorous parts of email management.
Claude does not send, reply to, or forward messages without consent by default. Users can change the approval settings for those actions, so the system can either pause for permission or run with fewer interruptions. That balance matters because drafting a reply is one thing; sending it to the wrong person is a very efficient way to create an afternoon.
Permission controls keep email actions visible
Individual users can toggle permissions for read-only tools and write/delete tools in the chat interface. Read-only tools are set to “Always allow” by default, while write/delete tools are set to “Needs approval.” This gives Claude access to inspect Gmail while keeping message changes and other destructive actions behind a confirmation step.
Team and enterprise plans add an administrative layer. Owners can decide whether members may let actions run without asking each time, which places approval rules under the organization’s control rather than leaving every user to configure them alone.
The setup points toward an AI assistant that handles email as a sequence of tasks instead of a single question-and-answer exchange. Claude can locate a message, read its contents, prepare a response, and send it when approved. It can also forward messages and maintain labels and threads, moving beyond search into inbox maintenance.
Google brings voice features out of beta
Google has also moved its AI-powered voice features for Gmail, Docs, and Keep out of beta. The features are now available to paying Google AI subscribers, following Gmail Live’s beta testing that began in June.
Gmail Live can answer questions about emails using natural language, then display transcripts and links. It is available on Android and iOS for Google AI Plus, Pro, and Ultra subscribers, giving users a voice interface for asking about information inside Gmail.
Docs Live takes a broader approach to document work. It can help organize thoughts and structure documents, accessing Drive, Chat, Gmail, and the web when permission is granted. Docs Live is available on Android and iOS for Google AI Pro and Ultra plans.
Keep Live handles smaller pieces of work but follows the same voice-first idea. It can write lists and notes from spoken input and understand implied instructions, rather than requiring every action to be stated as a rigid command. Keep Live is available on Android only for Google AI Pro and Ultra subscribers.
These updates divide the territory in a useful way. Claude’s Gmail connector focuses on delegated email actions with approval controls, while Google’s Live features focus on asking questions, organizing ideas, and creating content through voice. Both reduce the number of steps between an intention and an action, although each keeps access tied to permissions and subscription plans.
The meaningful change is not that AI can touch email, documents, or notes. That was always the easy demo. The important part is the control layer: Claude separates reading from writing and deletion, while Google limits its voice features by service, device, and Google AI plan.
Based on
