Anthropic last week announced its new platform, Claude controlled agentsAI for enterprises aims to cut out the more complex parts of agent deployment and competes with existing orchestration frameworks.
Claude Managed Agents is also an architectural change: enterprises already burdened with orchestrating an increasing number of agents can now choose to deploy orchestration logic. In the AI model layer.
While this comes with some potential advantages, such as speed (Anthropic offers its customers to deploy agents in days instead of weeks or months), it also, of course, gives the model provider — in this case, Anthropic — greater control over enterprise AI agent deployment and operations, potentially leading to greater results. "to lock" for the enterprise customer, making them more subject to Anthropic’s terms, conditions and subsequent platform changes.
But it might be worth it for your enterprise, as Anthropic further claims that its platform “manages complexity” and allows users to define agent tasks, tools and safeguards with a built-in orchestration harness, all without the need for sandboxing code execution, checkpointing, credential management, comprehensive permissions and end-to-end monitoring.
The framework manages state, execution schedules, and routing, and brings managed agents into the vendor-managed runtime loop.
Prior to the release of Claude Managed Agents, new direction VentureBeat research found that Anthropic is gaining traction at the orchestration level as enterprises adopt their native tools. Claude Managed Agents represents the firm’s new attempt to expand its footprint as the orchestration method of choice for organizations.
Interest in anthropic orchestras is growing
Orchestration has emerged as an important segment for enterprises scaling AI systems and deploying agent workflows.
A VentureBeat-focused survey of several dozen firms for the first quarter of 2026 found that enterprises are mostly choosing existing frameworks such as Microsoft’s Copilot Studio/Azure AI Studio, with 38.6% of respondents reporting using Microsoft’s platform in February. VentureBeat surveyed 56 organizations with more than 100 employees in January and 70 in February.
OpenAI followed closely with 25.7%. Both showed strong growth in the first two months of the year.
Over the past year, Anthropic has been struggling, fueled by increased interest in its offerings such as Claude Code.
Anthropic tools usage and workflow API adoption increased from 0% to 5.7% between January and February. This closely follows the growing adoption of Anthropic’s foundational models, and shows that enterprises using Claude are turning to the company’s native orchestration tools instead of adding a third-party framework.
While VentureBeat polled before the launch of Claude Managed Agents, we can extrapolate that the new tool will build on that growth, especially if it promises a simpler way to deploy agents.
Collapse of the outer layer of the orchestra
Enterprises may find simplified, internal integration for agents attractive, but it means giving up some control.
Session data is stored in a database controlled by Anthropic, which increases the risk of businesses being locked into a system operated by one company. This may be less desirable for some firms, and may compete with desires to move away from locked-down software-as-a-service (SaaS) applications in current stacks that many hope AI will facilitate.
The specter of vendor lock-in means that agent execution is model-driven rather than direct by the organization, occurring in an environment that enterprises do not fully control and making it difficult to guarantee behavior.
It also opens up the possibility of giving conflicting instructions to agents, especially if the only way users can exercise any control over agents is to address them with more context.
Agents can have two management plans: one defined by the enterprise orchestration system through instructions, and the other as a skill embedded from the Claude runtime.
This can be problematic for highly sensitive and regulated workflows such as financial analysis or customer-facing tasks.
Price, control and competitive kit
Balancing control with ease is one thing; businesses also consider the cost structure of Claude Managed Agents.
Presented by Claude Managed Agents hybrid pricing model combines token-based billing with usage-based uptime fees.
This makes managed Agents more dynamic but less predictable when setting cost structures. Businesses will be charged a standard rate of $0.08 per hour when agents are active.
For example, a one-hour session at $0.70 per hour can cost $37 to process 10,000 support tickets, depending on how long each agent works and how many steps they take to complete the task.
Microsoft, currently the leader in VentureBeat’s direction survey, offers several orchestration offerings. Copilot Studio uses a capacity-based billing structure, so businesses pay for blocks of interactions between users and agents rather than the number of steps an agent takes.
Microsoft’s approach is more predictable than Anthropic’s pricing plan: Copilot Studio starts at $200 per month for 25,000 messages.
Compared to similar competitors like OpenAI’s Agents SDK, the picture becomes bleak. The Agents SDK is technically free to use as an open source project. However, accounts for OpenAI basic API usage. For example, agents built and orchestrated with the Agents SDK using GPT-5.4 will cost $2.50 per 1 million input tokens and $15 per 1 million output tokens.
Enterprise decision
Claude Managed Agents provides respite to businesses that find the actual deployment of production agents too complicated. It reduces their engineering costs while adding speed and simplicity in a rapidly changing enterprise environment.
But it comes with a choice: losing control, traceability and portability, and risking additional vendor lock-in.
Anthropic simply explained why its ecosystem has become the primary model of choice for enterprises as well as orchestration infrastructure. It is becoming more important for businesses to balance ease with less control.
