just a few weeks Following the announcement of cloud managed agentsAnthropic has updated the platform with three new capabilities that collapse infrastructure layers like memory, evaluation, and multi-agent orchestration into a single runtime.
The move could threaten the standalone tools that many enterprises bundle together.
The new capabilities — ‘Dreaming,’ ‘Results,’ and ‘Multi-Agent Orchestration’ — are aimed at making agents inside cloud managed agents “more capable of handling complex tasks with minimal steering,” Anthropic said in a press release.
Dreaming is related to memory, where agents “reflect” on their many sessions and store memories so that they can learn and uncover unknown patterns. Results allow teams to define and set specific rubrics to measure an agent’s success, while multi-agent orchestration breaks down jobs so that a lead agent can delegate to other agents.
Cloud managed agents ideally provide enterprises with a simple path to deploy agents and embed orchestration logic in the model layer. It is an end-to-end platform to manage positioning, execution graphs, and routing. With Dreaming, Outcomes, and multi-agent orchestration, Cloud Managed Agents expands capabilities even further and competes directly with tools like LangGraph or CrewAI, as well as external assessment frameworks, RAG memory architectures, and QA loops.
threat of integration
Enterprises must now ask: Should we abandon our flexible, modular systems in favor of an agent platform that brings almost everything in-house?
Anthropic designed cloud managed agents to share context, state, and traceability in one place. This means the platform looks at individual decision agents, rather than requiring enterprises to stitch together disparate systems. It seems practical to have one platform that does everything. But not all enterprises want a full-service system.
Cloud managed agents already face criticism that it encourages vendor lock-in because it owns most of the architecture and tools that control agents. In the current paradigm, an organization can run managed agents but keep multi-agent orchestration, memory, or evaluation in a separate location to ensure resiliency.
The platform offers a fully hosted runtime, meaning that in-memory and orchestration runs on infrastructure that the enterprise does not own. This could become a compliance nightmare for some organizations that must prove data residency.
Another problem to consider is that enterprises already in the midst of massive AI transformations will have to cobble together workarounds to deal with the constraints of their technology stack. Not every workflow can be easily replaced by switching to cloud managed agents.
Dreaming and results against existing equipment
Most enterprises have a fragmented approach to AI deployment.
For example, they can use LangGraph or Crew AI for agent routing and workflow management, PineCone as a vector database for long-term memory, DeepEval for external evaluation, and human-in-the-loop quality assurance for reviewing certain tasks. Anthropic hopes to put an end to all that.
With Dreaming, Anthropic accesses memory by allowing users to actively rewrite it between sessions, so the agent essentially learns from its mistakes. Anthropic says this capability is useful for long-running states and orchestrations. Current systems often handle memory persistence by storing embeddings, retrieving relevant context, and adding more state over time.
The results address the evaluation part by describing the agents’ expectations. Instead of external quality checks, which are often performed by a team of humans, Anthropic is bringing evaluation into the orchestration layer instead of on top of it.
But it’s the multi-agent orchestration capability that makes cloud managed agents stand out against orchestration frameworks from Microsoft, Langchain, CrewAI, and others. Model providers like Anthropic and OpenAI have already begun to move aggressively in this area, arguing that bringing it down to the model level gives teams better control.
Big decisions will have to be taken
Enterprises face a big decision, and it may depend on where they are in agent maturity.
If an organization is still experimenting with agents and hasn’t deployed many agents in production, they may find it much easier to move to cloud managed agents and configure Dreaming and Results to their needs. This is the stage of development where, even if enterprises are using a third-party orchestrator like Langchain, they are still customizing it.
But for those who are already ahead in the process, the calculations become difficult. Now it is a matter of parallel evaluation and better understanding of their processes.
However, businesses will face the same decision even if they do not intend to use cloud managed agents. Anthropic has indicated that other models and platform providers will likely shift their product roadmaps to a similar model that keeps everything locked into a single system – because the models may be interchangeable, but the tooling and orchestration infrastructure will not.
<a href