Anthropic on Wednesday unveiled Claude Managed Agents in public beta, a cloud-hosted platform designed to dramatically accelerate how enterprises build and deploy AI-powered agents. The service handles the heavy lifting of infrastructure — hosting, scaling, monitoring, and error recovery — so engineering teams can focus on agent logic rather than plumbing.
From Months to Days
The core pitch is speed. According to Anthropic, the managed service compresses what has traditionally been a months-long development cycle into days or weeks. Rather than stitching together containers, state management, tool orchestration, and failure handling from scratch, developers define their agent's tasks, tools, and guardrails while Anthropic's infrastructure handles the rest.
Each agent runs in an automatically provisioned isolated container with built-in state management for data and credentials. The platform handles tool orchestration — determining which third-party applications an agent should call — and includes error recovery that lets agents resume seamlessly after outages.
Research Preview Features
Alongside the public beta, Anthropic is previewing two advanced capabilities. The first is a multi-agent feature that allows agents to spawn sub-agents for complex, multi-step tasks. The second is automatic prompt refinement, which Anthropic says improves task success rates by up to 10 points compared to standard prompting. For a wider look at the pattern, this guide on multi-agent AI systems shifting the enterprise from chatbots to operations covers how companies are architecting these systems in practice.
These research previews signal where the platform is headed: toward increasingly autonomous agent networks that can coordinate on enterprise-scale workflows.
Pricing and Access
Managed Agents bills based on underlying Claude model usage plus $0.08 per agent runtime hour. The service is available through the Anthropic Console, Claude Code, and a new CLI, giving developers multiple entry points depending on their workflow preferences.
Early Adopters Already Shipping
Notable early customers include Notion, Rakuten Group, and Asana. Several have already integrated agents into production, shipping use cases that span workflow automation, customer support copilots, and data operations agents.
The launch extends Anthropic's broader agent ecosystem, which now includes Claude Cowork for desktop and local file automation alongside Claude Code for agentic coding workflows. All reportedly share the same underlying agent loop and SDK patterns.
The Bigger Picture
The managed agents launch arrives as enterprise AI shifts from chatbot-style interfaces toward autonomous agents that can take real actions across business systems. For organizations evaluating where to begin, this guide on AI agents in the enterprise covers what to prioritize and what to avoid. OpenAI, Google, and Salesforce are all building competing platforms, but Anthropic is betting that a tightly integrated, infrastructure-managed approach will win developer loyalty.
With its revenue run rate reportedly reaching $30 billion and over 1,000 enterprise customers each spending more than $1 million annually, Anthropic has significant distribution to back the bet. Managed Agents represents the company's clearest play yet to own the full stack from model to deployment — and to make the leap from AI provider to enterprise platform.
