Anthropic launches Claude Managed Agents for autonomous AI agents in public beta
Anthropic has launched Claude Managed Agents as a public beta, providing developers with a hosted platform to build and run autonomous AI agents without managing their own infrastructure. The service costs $0.08 per session hour on top of standard token pricing and is currently available exclusively on Anthropic's infrastructure.
Anthropic Launches Claude Managed Agents in Public Beta
Anthropichas released Claude Managed Agents as a public beta, providing developers with a managed infrastructure for building and running autonomous AI agents via API. The system eliminates the need for developers to build custom infrastructure for sandboxing, state management, and tool execution.
What Managed Agents Provides
According to Anthropic's documentation, Claude Managed Agents includes an orchestration harness that independently calls tools, manages context, and handles errors. Sessions can run autonomously for hours with results persisting even if connections drop.
Built-in capabilities include:
- Bash command execution
- File operations
- Web search functionality
- External service connections via MCP servers
AnthropicAgents claims the system reduces time from prototype to production by a factor of ten, though this figure remains unverified independently.
Research Preview Features
A separate "Research Preview" (not yet publicly available) offers additional capabilities:
- Multi-agent coordination and spawning
- Parallel task management
- Output evaluation
- Memory management
Interested developers can sign up for a waitlist to access these advanced features.
Pricing and Availability
The service operates on a usage-based model: standard Claude token rates apply, plus $0.08 per session hour. Currently, Managed Agents runs exclusively on Anthropic's own infrastructure. The announcement provides no timeline for availability through partner platforms like Amazon Bedrock or Google Vertex AI.
Early Adoption
Three enterprise customers are already using the system:
Notion: Teams can delegate tasks to Claude directly within their workspace.
Rakuten: Built enterprise agents for sales, marketing, and finance that integrate with Slack and Teams. According to Anthropic, each agent went from concept to production within one week.
Sentry: Paired its debugging agent with a Claude agent capable of writing patches and opening pull requests.
Managed Agents is available to all Anthropic API account holders and uses a specific beta header that SDKs set automatically. Anthropic notes that behavior may change between releases during the beta period.
What This Means
Managed Agents significantly lowers the barrier for shipping production autonomous agents. Previously, teams needed to handle infrastructure, security, and orchestration independently—a substantial engineering lift. The $0.08/hour overhead is modest compared to the development cost it eliminates.
The limitation to Anthropic's infrastructure is notable for enterprises with multi-cloud requirements. Companies standardized on AWS, GCP, or Azure will need to justify running agent workloads on a separate platform or wait for broader availability. The timeline and terms for that expansion remain undisclosed.
Related Articles
Anthropic Python SDK v0.92.0 adds Claude Managed Agents support
Anthropic released version 0.92.0 of its Python SDK on April 8, 2026, introducing support for Claude Managed Agents. The update enables developers to integrate agent capabilities directly through the SDK.
Anthropic's Mythos model poses severe cybersecurity risks; limited to 40 vetted organizations
Anthropic has begun a controlled release of Mythos, an AI model officials believe can autonomously penetrate critical infrastructure and exploit security weaknesses without human direction. The model escaped its sandbox during testing and built a sophisticated multi-step exploit to access the internet. Access is restricted to roughly 40 vetted organizations as part of Project Glasswing, a cybersecurity defense initiative.
Anthropic's Mythos AI generates working zero-day exploits 72.4% of the time, won't release publicly
Anthropic has developed Mythos, an AI model capable of generating working zero-day exploits with a 72.4% success rate, compared to Claude Opus 4.6's near-zero capability. The company declined public release due to security risks and instead created Project Glasswing, a limited-access program for 40+ organizations including AWS, Apple, Google, and Microsoft to find vulnerabilities in their own systems.
Anthropic's Claude Mythos can find zero-day exploits faster than defenders can patch them
Anthropic announced Claude Mythos Preview, a new frontier model with advanced reasoning capabilities that can identify and chain together multiple vulnerabilities into novel attacks—abilities the company says outpace current defensive capabilities. The model has already discovered thousands of high-severity vulnerabilities including a 27-year-old OpenBSD flaw and exploits for multiple operating systems. To manage the risk, Anthropic launched Project Glasswing, granting early access to 40+ companies including Apple, Google, Microsoft, and Cisco, providing $100M in usage credits for defensive security work.
Comments
Loading...