Skip to content
Trending storyDeveloping

Anthropic decouples agent architecture from Claude delivery

2 reports1 sourceupdated 2 hours ago

What happened

AI digest

On October 28, Latent Space published an interview with Katelyn Lesse and Angela Jiang on how the Claude Platform separates the model, execution framework and sandbox. It covers the Opus 5.5 release, API expansion, isolation, credential protection, network controls and keeping task progress across context windows. A newly recorded intro adds Anthropic's S-1, its improvement-and-investment loop and the platform's significance. Later reporting says Anthropic builds its own products on the same platform components. Managed Agents split the session log, runtime and code execution sandbox, so a sandbox failure need not end the whole session; service credentials are stored separately and attached through a proxy. The Opus 5.5 release also involves API integration changes against knowledge distillation, and customers evaluating it in their own runtimes and reporting anomalous behavior.

Written by AI from the coverage · updated 2 hours ago

Coverage

Follow the reports to see the story from different sides.

Oct 28
  1. Latent SpacePick
    How Anthropic ships Claude with Claude: splitting an agent's brain from its hands

    Anthropic's engineering and product leads explain how they build their own products on shared Claude Platform components, and how they coordinate model releases, safety and scaling. Managed Agents separate session logs and the runtime framework from the code execution sandbox, so a sandbox failure doesn't kill the whole session. Service credentials are stored separately and attached through a proxy. The Opus 5.5 release also involved API integration changes to prevent knowledge distillation, and customers running evals in their own frameworks fed back anomalous behavior.

  2. Latent Space
    How Anthropic ships the #1 LLM in the World safely: Separating the Brain from the Hands

    Katelyn Lesse 与 Angela Jiang 在 Latent Space 访谈中讲解 Claude Platform 的工程设计,涵盖 Opus 5.5 发布、API 扩展和可靠托管智能体的构建。访谈重点包括模型、执行框架与沙箱的解耦,以及隔离、凭据保护、网络控制和跨上下文窗口维持任务进度。新录制的导言补充了 Anthropic 的 S-1、改进与投资循环及平台意义。

Heat over time

Not enough continuous observations to draw a trend yet.