Anthropic Revolutionizes AI with Its Autonomous Agents Claude

Anthropic Unveils Its Autonomous AI Agents Claude

Anthropic has recently introduced the "Claude Managed Agents," a public beta solution that allows developers to design autonomous AI agents without having to manage their own infrastructure. This offering, accessible via an API, aims to simplify the process of creating and managing AI agents by providing a hosted platform.

Anthropic's Claude Managed Agents offer a hosted platform that enables developers to build and run autonomous AI agents. Notion, Rakuten, and Sentry are among the first users of this system. These companies are using the agents to automate tasks in the workplace, integrate agents into platforms like Slack, and even for automated debugging.

A Turnkey Solution for Developers

The system operates exclusively on Anthropic's infrastructure and offers a pricing model of $0.08 per session hour, in addition to standard token rates. The Claude Managed Agents provide an orchestration framework that automatically manages tools, context, and errors, reducing the time required to move from prototype to production by a factor of ten.

Sessions can be autonomous for several hours, with results persisting even in the event of a connection drop. Built-in tools include bash commands, file operations, web searches, and connections to external services via MCP servers.

A distinct "Research Preview," which is not yet publicly available, allows agents to create other agents, coordinate parallel tasks, evaluate outputs, and manage memory. Interested individuals can sign up for a waitlist to access these features.

Companies Already Enthralled by the Solution

Notion is using this technology to enable teams to delegate tasks directly within their workspace. Rakuten has developed agents for sales, marketing, and finance, integrated into Slack and Teams, and operational within a week. Sentry, for its part, has paired its debugging agent with a Claude agent to automate the writing of patches.

The Managed Agents are available for all API accounts and require a specific beta header that the SDK automatically defines. Anthropic notes that behavior may change between versions. Currently, the system is limited to Anthropic's infrastructure, with no mention of future availability through partner platforms like Amazon Bedrock or Google Vertex AI. This exclusivity could pose challenges for companies with multi-cloud strategies.