Anthropic released a new collaborative tool for its artificial intelligence (AI) chatbot Claude on Tuesday. The tool will allow users to ground the output of the chatbot to the internal knowledge provided to it.

In a newsroom post, Anthropic said that Projects is designed to bring together curated sets of knowledge and chat activity in a single interface. Projects are also shareable, so the entire organisation or a specific team can view the best chats with the AI. The AI firm also said that the feature will enable the AI chatbot to generate useful ideas, make more strategic decisions, and show results that are aligned with the user’s goals.

Projects are similar to AI agents such as Google Gems or OpenAI’s GPTs.

Users creating a project can upload multiple documents to create an internal knowledge base. The AI will then curate responses entirely based on this knowledge (alongside its pre-training data), and share responses that are more relevant. To further tailor the responses, users can add custom instructions for each project.The company said each project will have a 2,00,000 token context window, which is close to a 500-page book. This means the AI can contextualise the information across a wide number of style guides, codebases, interview transcripts, or past work shared with it. The feature also runs on the recently released Claude 3.5 Sonnet, which Anthropic said outperforms Claude 3 Opus, GPT-4o, and Gemini 1.5 Pro.

The main benefit of Projects over other AI agents is that it is also shareable. That means users can share the best chats with Claude with other teammates in a centralised activity feed. The company says this will allow all teammates to gain insights from a project. Projects are available on Claude.ai for all Pro and Team subscribers. Notably, the Claude Pro subscription is priced at $20 (roughly Rs. 1,670) per person per month and the Claude Team subscription costs $25 (roughly Rs. 2,090) per person per month with a minimum of five members added.

visit:gamicaltech..com