Section 1
What Just Happened
On April 8, 2026, Anthropic launched Claude Managed Agents. It's a suite of APIs that lets you build, deploy, and run AI agents at scale on Anthropic's own infrastructure.
Before this, building a production agent meant months of work. You needed sandboxed code execution, state management, credential handling, scoped permissions, error recovery, and end-to-end tracing. Most teams spent more time on infrastructure than on the agent itself.
Managed Agents handles all of that. You define your agent's tasks, tools, and guardrails. Anthropic runs it. The built-in orchestration harness decides when to call tools, how to manage context, and how to recover from errors.
The result: teams are shipping agents in days instead of months. Notion, Asana, Sentry, Rakuten, and Atlassian are already live with it.
Section 2
What You Get Out of the Box
Managed Agents is not just an API wrapper. It's production infrastructure you'd normally spend months building yourself. Here's what's included:
Production-Grade Agents
Secure sandboxing, authentication, and tool execution handled for you. Your agent runs in an isolated environment with proper security built in from the start.
Long-Running Sessions
Agents can operate autonomously for hours. Progress and outputs persist even through disconnections. You don't need to babysit them.
Multi-Agent Coordination
Agents can spin up and direct other agents to parallelize complex work. One agent breaks down a task, assigns sub-agents, and combines results. Currently in research preview.
Trusted Governance
Scoped permissions, identity management, and execution tracing. You control exactly what your agent can access. Every tool call and decision is logged and inspectable in the Claude Console.
In internal testing, Managed Agents improved task success by up to 10 points over a standard prompting loop. The biggest gains were on the hardest problems.
Section 3
Who's Already Using It
This isn't vaporware. Major companies shipped with Managed Agents before it even went public beta:
Notion
Teams delegate work to Claude directly inside their workspace. Engineers ship code, knowledge workers produce websites and presentations. Dozens of tasks run in parallel while the team collaborates on output.
Sentry
Their debugging agent Seer flags a bug, a Claude-powered agent writes the patch and opens the PR. Developers go from flagged bug to reviewable fix in one flow. The integration shipped in weeks.
Asana
Built AI Teammates that work alongside humans inside Asana projects. They take on tasks, draft deliverables, and collaborate in real-time. Built with Managed Agents, shipped dramatically faster than building from scratch.
Rakuten
Deployed specialist agents across product, sales, marketing, and finance. Employees assign tasks through Slack and Teams and get back deliverables like spreadsheets, slides, and apps. Each agent was deployed within a week.
Atlassian (Jira)
Built agents for developers directly into Jira workflows. Teams assign tasks right from Jira. Managed Agents handles sandboxing, sessions, and scoped permissions so their engineers focus on features, not infrastructure.
Section 4
What You Can Build
Managed Agents is flexible enough for single-task runners and complex multi-agent pipelines. Here are the categories of agents teams are building right now:
Coding Agents
Read a codebase, plan a fix, write the code, and open a pull request. From bug report to reviewable PR without a human touching the code.
Productivity Agents
Join a project, pick up tasks, and deliver work alongside the rest of the team. Think of it as an AI teammate that actually ships.
Finance and Legal Agents
Process documents, extract what matters, answer queries on the fly. One team built an agent that dynamically creates its own tools to handle any question about their document corpus.
Meeting Prep Agents
Research every participant ahead of a meeting. Surface what matters for moving the conversation forward. Connects to your calendar, CRM, and meeting notes automatically.
Section 5
How to Set Up Your First Agent
Getting started is simpler than you'd expect. There are three ways in:
Claude Console (Easiest)
Go to platform.claude.com and navigate to the Agent Quickstart. Walk through the guided setup. Define your agent's task, attach tools, set permissions, and deploy. No code required to get started.
Claude Code CLI
If you already use Claude Code, you can build with Managed Agents right from your terminal. Just type "start onboarding for managed agents in Claude API" and the built-in claude-api skill walks you through the setup.
API Direct
For developers who want full control, the Managed Agents API lets you create sessions, define tools, and manage agent lifecycles programmatically. Full docs at platform.claude.com/docs/en/managed-agents/overview.
Quick start -- Claude Code
$ claude
> start onboarding for managed agents in Claude API
The Claude Console gives you session tracing, integration analytics, and troubleshooting guidance. You can inspect every tool call, decision, and failure mode your agent makes.
Section 6
What It Costs
Managed Agents is priced on consumption. You pay standard Claude Platform token rates plus $0.08 per session-hour for active runtime.
That's it. No setup fees, no minimum commits, no enterprise-only pricing wall. You pay for what your agents actually use.
For context: if your agent runs for 10 minutes to process a batch of documents, that's about $0.01 in runtime plus your normal token costs. A coding agent that takes 30 minutes to write and test a fix costs about $0.04 in runtime.
The infrastructure this replaces (sandboxing, state management, auth, tracing) would normally cost you a full engineering team for months. That's the real math.
Section 7
Why This Matters
Hundreds of startups spent the last two years building agent infrastructure platforms. Secure sandboxing, session management, tool orchestration, credential handling. That was the product.
Anthropic just made all of it a built-in feature on their platform. For $0.08 an hour.
This is the pattern we've seen over and over in AI: third-party tools solve a real problem, then the foundation model company ships it as a native feature. MCP did it for tool connections. Managed Agents does it for agent deployment.
If you're building with AI, this is the inflection point where "building an agent" stops being a six-month infrastructure project and becomes a weekend experiment. The barrier just dropped by an order of magnitude.
The teams that move first will have production agents running while everyone else is still reading the docs.
Zero to Automated
Get the full AI operating system
Guides are a great start. But inside Zero to Automated, you get a done-for-you AI system deployed within 48 hours, plus everything you need to keep building:
- Claude Code 101 — 8 lessons, zero to automating real work by Sunday
- Ready-to-install Claude Skills — a growing library with 2+ new skills every week
- Done-for-you workflows — copy, paste, run. New ones every week.
- Weekly live builds — watch real projects get built, ask questions, follow along
- Small group coaching — direct access to both founders
Price locked for life at whatever you join at.