Claude Managed Agents and Smart Reports: Anthropic's Enterprise Agent Stack

#anthropic #claude #managedagent #enterprise

TL;DR — Anthropic's Claude Managed Agents, in public beta since April, abstracts away the infrastructure of running production agents — sandboxing, authentication, checkpointing, and long-running sessions — at $0.08 per runtime hour. On September 10, the company layered on Smart Reports, an analytics tool that tells enterprises what their agents actually cost, where they stall, and which workflows are worth packaging as shared skills. Together they form Anthropic's enterprise agent stack.

Introduction

Building a production AI agent has always meant two jobs. The first is designing what the agent does. The second is building everything that makes it run: sandboxed execution, state management, credential handling, error recovery, tool orchestration, and checkpointing. That second job took most teams three to six months — and had nothing to do with the agent itself (Source : The AI Corner — Claude Managed Agents: complete guide).

Anthropic's Managed Agents product eliminates the second job. You define the tasks, tools, and guardrails; Anthropic runs the infrastructure. The reaction was immediate — one developer's "There goes a whole YC batch" post pulled two million views in two hours.

Managed Agents: The Infrastructure Layer

The public beta, launched April 8, 2026, bundles secure sandboxed code execution, authentication, checkpointing, and scoped permissions. Sessions are persistent and long-running, surviving disconnections and resuming exactly where they stopped. Built-in tool orchestration and automatic error recovery handle the failure modes that normally consume engineering time, while session tracing in the Claude Console gives full visibility into every agent action.

Pricing is the part that reshapes the build-vs-buy calculus. Runtime costs $0.08 per hour plus standard Claude model usage, meaning an agent running around the clock costs about $58 per month in runtime alone, before token costs. Multi-agent coordination — agents that spin up other agents — and self-evaluation are both in research preview (Source : The AI Corner — Claude Managed Agents: complete guide).

Early Deployments

The early-customer list reads like a who's who of enterprise SaaS. Notion lets teams delegate coding, slides, and spreadsheets to Claude without leaving the workspace, running dozens of parallel tasks. Rakuten deployed specialist agents across product, sales, marketing, finance, and HR, each live in under a week. Asana built AI Teammates that pick up assigned tasks inside projects, and Sentry built an agent that goes from a flagged bug to an open pull request fully autonomously. Vibecode reports users spin up the same infrastructure at least 10x faster than before.

Smart Reports: The Accountability Layer

Smart Reports, launched September 10 in beta on Claude Enterprise plans, closes the loop that Managed Agents opened. It analyzes how a team uses Claude and reports on the work getting done, what it costs, where sessions run into friction, and which repeated patterns are worth packaging as shared skills (Source : Claude Help Center — Release notes).

Each report breaks usage into workstreams, deliverables produced, cost per session by output type, task outcomes, and most common frictions — a connector that was never set up, output that didn't match the ask, approval gating, or rework loops. It also surfaces reusable skills and workflows to build. During beta, organizations can run up to 10 reports per month for free, with the limit resetting monthly.

One constraint is explicit and worth noting: Smart Reports are designed to guide adoption and investment, not to evaluate individual performance or make employment decisions (Source : Claude Help Center — Get started with smart reports).

Why This Matters

Managed Agents and Smart Reports are two halves of the same thesis: agents become a line item you can run and measure, not a project you staff. The infrastructure layer removes the build cost; the analytics layer surfaces the unit economics. For an enterprise deciding whether autonomous agents are real or hype, that combination is more persuasive than any benchmark.

FAQ

What does Claude Managed Agents cost?
$0.08 per runtime hour plus standard Claude model usage. A 24/7 agent costs roughly $58 per month in runtime before tokens.

What is Smart Reports?
A beta analytics tool on Enterprise plans that reports what Claude usage costs, where it stalls, and which patterns are worth packaging as shared skills.

Who is using Managed Agents?
Notion, Rakuten, Asana, Sentry, and Vibecode are among the early deployments.

Is this enterprise-only?
Managed Agents is in public beta for developers broadly; Smart Reports is gated to Claude Enterprise plans.

DEV Community