Anthropic has launched Claude Managed Agents, a cloud-hosted platform designed to let companies build and deploy autonomous AI agents without building the underlying infrastructure themselves, as the company disclosed that its annualised recurring revenue has surpassed $30 billion, roughly triple the figure reported in December 2025.
The product, now in public beta, addresses a persistent bottleneck in enterprise AI adoption: most businesses that want to run production-grade agents spend months engineering sandboxed execution environments, state management systems, credential handling and permission controls before writing a single line of user-facing logic.
Managed Agents abstracts that work away entirely, supplying a pre-built agent harness, a sandboxed environment for secure software projects, and built-in orchestration that handles tool calling, context management and error recovery out of the box.
Agents can run autonomously for extended periods, with sessions persisting through disconnections, and administrators can monitor peer agents and toggle tool permissions from a central dashboard inside the Claude Console, which logs every tool call, decision point and failure mode for debugging purposes.
Angela Jiang, head of product for the Claude Platform, said the offering enables any business to deploy a fleet of Claude agents without needing to assemble bespoke distributed-systems engineering internally.
Internal testing showed the platform improved success rates on structured file generation tasks by up to 10 percentage points compared with standard prompting approaches, with the gap widening on more complex problems.
Early adopters span a range of industries and use cases.
Notion, the productivity software company, has integrated Managed Agents directly into its workspaces, allowing engineers to ship code and knowledge workers to generate presentations and websites in parallel, with the system handling dozens of simultaneous tasks.
Rakuten, the Japanese e-commerce conglomerate, deployed agents across product, sales, marketing, finance and HR functions within a week per deployment, with agents accepting task assignments via Slack and Microsoft Teams and returning deliverables, including spreadsheets and slide decks.
Related reading
- Anthropic spots 'emotion vectors' inside Claude
- Memento-Skills lets AI agents rewrite their own capabilities without retraining underlying models
- Microsoft warns AI agents risk becoming "double agents" as it unveils security controls at RSAC
Sentry, the software monitoring company, paired Managed Agents with its existing debugging tool to create a pipeline in which a flagged software bug flows automatically to a reviewable code fix, a workflow the company said took weeks to ship rather than the months originally projected.
The launch comes as Anthropic competes directly with OpenAI, Google and Salesforce for enterprise agent infrastructure, with all four companies moving aggressively to embed their models in long-term production workflows ahead of anticipated initial public offerings this year.
The recap
- Anthropic launches Claude Managed Agents to deploy AI agents.
- Annualized recurring revenue surpasses $30 billion, threefold since December 2025.
- Anthropic and OpenAI are preparing to go public this year.
