# Why both vendors chose the same design (/claude-and-chatgpt-2/architecture/why-both-vendors-chose-the-same-design) --- type: Document title: "Why both vendors chose the same design" description: "What each feature is for, and the seven constraints that led Anthropic and OpenAI to similar designs." status: stable order: 0.42 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-11-09T20:42:36Z chapter: "F00A" part: Front expert_status: required concepts: [ "Why the same design" ] last_verified: 2026-11-09 sources: - id: cowork-architecture title: "Claude Cowork architecture overview (Anthropic, verified 2026-11-09)" resource: https://support.claude.com/en/articles/14479288-claude-cowork-architecture-overview - id: codex-harness title: "Unlocking the Codex harness: how we built the App Server (OpenAI, February 2026, verified 2026-11-09)" resource: https://openai.com/index/unlocking-the-codex-harness/ - id: cloud-scheduled-tasks title: "Automate work with scheduled tasks (Anthropic, verified 2026-11-09)" resource: https://code.claude.com/docs/en/routines - id: desktop-scheduled-tasks title: "Schedule recurring tasks in Claude Code Desktop (Anthropic, verified 2026-11-09)" resource: https://code.claude.com/docs/en/desktop-scheduled-tasks - id: agents-api title: "Agents API architecture (OpenAI, verified 2026-11-09)" resource: https://developers.openai.com/api/docs/guides/agents-api/architecture - id: work-cloud-security title: "ChatGPT Work cloud security (OpenAI, verified 2026-11-09)" resource: https://learn.chatgpt.com/docs/enterprise/chatgpt-work-cloud-security - id: new-stack title: "OpenAI is folding Codex into the ChatGPT app (The New Stack, July 2026, verified 2026-11-09)" resource: https://thenewstack.io/openai-codex-work-atlas/ generated: at: 2026-11-09T20:42:36Z by: process:panaversity trust_tier: unverified build_id: sha256:46fd53c5d7ab1c14f6347908d45c9fdaecda4d914439923ee878bdcecd57803e dirty: true ksor_version: 0.0.60 --- ### The goals These features let an AI agent do real work on a computer. They let it keep working after you close the tab. And they let it do this safely, while the model itself stays a stateless service. Anthropic and OpenAI face similar constraints and made similar choices. Both put a harness in an isolated cloud sandbox around a stateless model, and store lasting state elsewhere. This is one design pattern, not the only one. Local and self-hosted agents arrange things differently. ### What each feature is for | Feature or change | Goal | Basis | | --- | --- | --- | | A cloud sandbox for each session (a VM or a container) | Give the agent a real computer (a shell, files and code) without touching your machine or other users' work | Isolation for each session is vendor-documented.[^cowork-architecture] The goal is inferred | | The harness runs in the cloud, not in your browser or app | Work continues when the tab closes or you switch devices | Vendor-documented. OpenAI says keeping state on the server "means work continues even if the tab disappears"[^codex-harness] | | One harness behind every surface | The same agent behaves the same way on the web, desktop, mobile, the CLI and the IDE (the code editor) | Vendor-documented for Codex: "the same Codex harness" powers all experiences. For Claude, inferred: the session examined was a chat running on Claude Code, and the cloud docs list the same environments for every surface | | Cloud scheduled tasks | Unattended work that recurs or starts on an event, and runs with your computer off | Vendor-documented. Tasks "keep working when your laptop is closed"[^cloud-scheduled-tasks] | | The bridge to your computer, and Desktop tasks | Reach local files and apps only when a task needs them, inside folders you choose | Vendor-documented[^cowork-architecture],[^desktop-scheduled-tasks] | | An egress proxy, and connector tokens kept outside the sandbox | Limit the damage if untrusted content steers the agent | Vendor-documented[^cowork-architecture] | | The conversation is stored on the backend, and the sandbox can be thrown away | Resume from anywhere, and pause or reclaim machines cheaply | Inferred | Two terms in the table are new. An **egress proxy** is a gatekeeper that all traffic leaving the sandbox must pass through. A **connector token** is the key that lets a connector sign in to one of your apps. ### Seven shared constraints, similar design choices **[Inferred]** This is our reasoning from the documented features. Neither vendor states it. ![Seven shared constraints, each leading to a design response: a sandboxed computer, a harness loop, a server-side harness, one harness behind every app, reuse of Claude Code and Codex, disposable VMs with state outside, and keys outside with approvals required. Together they give the common pattern: UI, backend, a harness in an isolated sandbox, then a stateless model, with state and authority outside the sandbox.](img/seven-shared-constraints.png) 1. **Agents must act, not just answer.** Useful work means calling tools. An agent that only calls outside services may need no computer, as OpenAI's Agents API allows.[^agents-api] But code and files need a computer. Because the agent can be wrong or steered, that computer is isolated. 2. **The model is stateless, and its context is limited.** Something must build each call's context, run the tools, compact the context and track progress. That is the harness. 3. **Work outlives the screen.** A loop in a browser tab stops when the tab closes. Long, scheduled or cross-device work needs the harness on the server. 4. **The same agent should appear everywhere.** One harness behind every app is cheaper and more consistent than one harness for each app. 5. **Coding agents had shown that the pattern works.** Claude Code and Codex already combined a loop, tools and a sandbox. OpenAI documents that ChatGPT Work's cloud tasks use the Codex harness.[^work-cloud-security] In the Claude session examined, a chat ran on Claude Code. Press coverage reports the same direction.[^new-stack] 6. **Fresh machines are cheap and safe.** Light VMs start quickly, so a fresh one per session is affordable. Throwing them away limits leaks and frees capacity. So lasting state must live outside the machine. 7. **Safety has to sit outside the agent.** The agent can be wrong or manipulated. So both vendors keep credentials out of the sandbox, filter network traffic and require approvals. See [Security boundaries](security-boundaries.md). The common pattern: UI → backend → harness in an isolated sandbox → stateless model. Lasting state and authority (the power to decide what is allowed) stay outside the sandbox. Claude and Codex are two versions of it. [^cowork-architecture]: As verified on November 9, 2026. Claude Cowork architecture overview, Anthropic. [^codex-harness]: As verified on November 9, 2026. Unlocking the Codex harness: how we built the App Server, OpenAI. [^cloud-scheduled-tasks]: Automate work with scheduled tasks, Anthropic. [^agents-api]: Agents API architecture, OpenAI. [^work-cloud-security]: As verified on November 9, 2026, OpenAI says "Work Cloud uses the Codex task-execution harness". ChatGPT Work cloud security, OpenAI. [^new-stack]: OpenAI is folding Codex into the ChatGPT app, The New Stack. [^desktop-scheduled-tasks]: As verified on November 9, 2026. Schedule recurring tasks in Claude Code Desktop, Anthropic.