Anthropic announced on Tuesday that its AI chatbot, Claude, now integrates with Google Workspace, allowing it to search and reference your emails in Gmail, scheduled events in Google Calendar, and documents in Google Docs.
The integration is rolling out in beta first to subscribers to Anthropic’s Max, Team, Enterprise, and Pro plans. Administrators managing multi-user accounts must enable the integration on their end before users can connect their Google Workspace and Claude accounts, according to Anthropic.
Google DeepMind’s Gemini chatbot also integrates with Workspace, and OpenAI’s ChatGPT integrates with Google Drive. However, Anthropic is one of the first third-party AI companies to offer a way to closely connect to Google’s productivity suite.
Anthropic’s team-up with Google aims to give Claude more personally tailored responses without requiring users to repeatedly upload files or craft detailed prompts. OpenAI and Google have tried achieving the same effect via different approaches, such as adding memory features that allow chatbots to reference past conversations in their replies.
In a press release, Anthropic says Claude’s new integration can help users organize their professional and personal lives. For example, Anthropic claims the feature can assist parents by scanning “emails and calendar events to highlight important commitments, while searching the web for updated school calendars, local community events, and weather forecasts that might affect family plans.”
Claude will provide in-line citations when it references Workspace content, showing users exactly where specific information originated, says Anthropic.
While the integration doesn’t give Claude the ability to schedule calendar events or send emails, it may raise security concerns among some users. It’s unclear how extensively Claude will search through a person’s Google Workspace, or whether users have to direct Claude to look at a particular email or calendar event depending on the nature of their request. It’s also not clear whether users can ask Claude not to search across certain sensitive emails or files.
Responding to the above, an Anthropic spokesperson told TechCrunch that the company doesn’t train models on user data by default and has implemented “strict authentication and access control mechanisms” for external services like Workspace.
“Each user or organization’s connections to external services (like Google Drive, Gmail, etc.) are properly authenticated and authorized for only that specific user or organization,” the spokesperson said in a statement. “Claude doesn’t have the ability to access or transfer data between different users’ connected services, as each connection is bound to the specific authentication credentials of that individual user or organization.”
Anthropic also announced on Tuesday the launch of Claude Research, a new feature that conducts multiple web searches to generate detailed answers. Positioned as a competitor to OpenAI and Google’s “deep research” agents, Claude Research offers an “optimal tradeoff” between speed and comprehensiveness, Anthropic says.
Claude Research typically runs for less than a minute to compile info, according to an Anthropic spokesperson — faster than some rival deep research agents. However, Claude Research doesn’t use a custom model, instead leveraging Claude’s recently launched web search capabilities.
The company is rolling out Claude Research to subscribers to its Max, Team, and Enterprise plans in the United States, Japan, and Brazil. It’ll come to Pro customers soon, Anthropic says.
These updates are part of Anthropic’s broader effort to attract users to its AI subscription plans with features that make Claude more capable and useful. While Claude is growing in popularity, reaching 3.3 million web users in March, according to data compiled by SimilarWeb, Anthropic’s user base is still dwarfed by ChatGPT’s.
Add comment