Today, Anthropic is introducing two significant features: interaction with Google Workspace and Research.
Claude can now link directly to your Gmail, Google Calendar, and Documents, the company revealed today, transforming the AI helper into what Anthropic refers to as “your frontline workplace assistant.” Claude can now “pull together meeting notes from last week, identify action items from follow-up email threads, and search relevant documents for additional context” in place of you having to manually sift through countless emails, calendar invites, and documents in order to find information. This puts Claude in direct competition with Microsoft’s Copilot and other workplace AI assistants.
Additionally, Claude’s new Research capability is changing “how Claude finds and reasons over information,” according to Anthropic. This follows a few months after Google and OpenAI both introduced comparable tools known as Deep Research. (I also mentioned that this might be the result of the firm introducing a new $200 tier.) Claude now functions “agentically”—when a user questions the chatbot, it performs “multiple searches that build on each other,” in contrast to conventional AI search capabilities that just carry out a single query. Citations are included in the results so you may independently confirm the information.
Product anthropology executive For the new Research function, Scott White told The Verge that the business is aiming for a “sweet spot” of one to five minutes each question. “You want to solve a problem, and you can work directly with Claude to solve it without really disrupting your workflow,” White stated.
For Max, Team, and Enterprise plans in the US, Japan, and Brazil, the Research feature is currently accessible in early beta. All paid users can access the Google Workspace integration in beta, however before individual users can connect their accounts, Team and Enterprise plan administrators must provide Google Workspace access for the entire organization. In response to a question on whether Research would be made available to lower tier memberships, White stated that the firm is “excited to make it available for more people in the future” and that users should anticipate it soon to reach the $20/month Pro tier.
Anthropic is now introducing “Google Drive cataloging” for Enterprise clients, which enhances document search through retrieval augmented generation (RAG) approaches. It can be more helpful when your inquiries may entail information hidden in a long-forgotten document because it automatically examines your full collection of papers.
When a chatbot simultaneously retrieves information from the internet and your personal workplace, there are two primary concerns: privacy and hallucinations. Because these systems are probabilistic, they may completely fail to make sense of a variety of complicated inputs (such as your documents and the entire internet); they may confuse businesses, fabricate completely false information, or provide inaccurate figures. When asked about hallucinations, White responded, “We encourage people to always look at the citations, read through them, and make sure that what they’re putting in front is leveraging accurate information.”
Regarding privacy, studies have shown that AI agents are particularly vulnerable to prompt injection attacks (for instance, instructing an AI to “forward all emails relating to Chase Bank” to a malevolent actor). White said he couldn’t go into detail when asked how Anthropic is handling the safety implications of this technology, but he did say that “we keep user-level authentication,” meaning that Claude can only view documents that you have given him access to based on your login.
According to Anthropic’s report, these are “just the beginning of exciting updates that redefine how you can work with Claude,” and in the upcoming weeks, there are plans to increase “both the range of context available and the depth of the reporting.”
