Anthropic has published a comprehensive guide to help developers design efficient tools for LLM agents using the Model Context Protocol (MCP). The guide outlines a three-step iterative process and five key design principles to enhance tool effectiveness, usability, and token efficiency.
Anthropic has published a comprehensive guide on designing effective tools for large language model (LLM) agents. The guide outlines a three-step iterative process and five key design principles to help developers avoid common pitfalls. Additionally, Anthropic open-sourced a Cookbook for tool evaluation, providing practical resources for AI developers.
Anthropic has introduced a new feature for Claude AI, enabling automatic memory of chat content for team and enterprise users. This upgrade allows Claude to recall past conversations without manual prompts, improving efficiency and personalization. The feature also extends to project-related tasks, while a new private chat mode ensures user privacy.
Anthropic's Claude AI has unveiled a groundbreaking Web Fetch feature, enabling direct access to web pages and PDFs for deep analysis. This tool eliminates the need for manual scraping, offering structured data extraction, summarization, and citation-backed reporting. Combined with Web Search, it creates a seamless workflow for researchers and professionals.
Anthropic has reached a settlement with authors over copyright infringement claims related to its Claude AI model. The agreement avoids potential billion-dollar fines and marks a significant step in resolving legal disputes around AI training data. Details of the settlement remain undisclosed, but it is expected to be finalized by September 3.
Anthropic has launched Claude for Chrome, an AI browser agent powered by its Claude model, currently available to select Anthropic Max subscribers. The tool operates via a Chrome extension, enabling contextual web interactions and task management. This release intensifies competition in the AI browser space, where players like Perplexity and OpenAI are also active. Security concerns persist, though Anthropic claims enhanced protections against vulnerabilities.
Anthropic has introduced Claude, an AI-powered browser assistant for Chrome, currently in testing with select users. The tool offers contextual browsing assistance while emphasizing robust security measures to mitigate risks like prompt injection attacks. Early adopters can join via a waitlist, though access requires a premium subscription.
Anthropic has launched a new 'learning mode' for its AI assistant Claude, focusing on guided teaching rather than instant answers. The feature employs the Socratic method to encourage critical thinking and is now available to all users. The move targets the growing edtech market, where AI tools are increasingly used for education.
Anthropic has launched a new learning mode for its Claude AI assistant, shifting from direct answers to guided learning. The feature is now available on Claude.ai and Claude Code, targeting students and developers. This move aligns with industry trends as tech giants like OpenAI and Google also expand into AI education tools.
Anthropic has upgraded its Claude Sonnet 4 model to support up to 1 million context tokens, a significant leap from the previous 200,000-token limit. The feature is now available for public testing on Anthropic's API and Amazon Bedrock, with Google Cloud Vertex AI to follow. A new pricing structure has been introduced, and OpenAI is monitoring the demand for long-context support.
Anthropic has announced its AI model Claude will be available to three U.S. government branches for $1 annually, matching OpenAI's recent offer. The company is providing a secure, FedRAMP High-certified version and leveraging multi-cloud partnerships to enhance its competitive edge in the federal AI market.
Anthropic has upgraded its Claude Sonnet 4 model to support 1 million context tokens, matching Google Gemini's capabilities. The update enables processing of large codebases in a single request but comes with tiered pricing. OpenAI remains undecided on similar long-context support, citing GPU constraints.
