Claude Code can now run entirely on your own infrastructure — a beta aimed squarely at regulated industries

Anthropic opened a public beta on 6 August 2026 letting Claude Code Team and Enterprise customers run coding-agent sessions on servers inside their own network via a single 'claude self-hosted-runner' command, keeping repository checkouts, build artifacts and secrets off Anthropic's infrastructure entirely — a direct answer to the data-residency objections that have kept healthcare and other regulated sectors from adopting AI coding agents.

Anthropic opened a public beta on 6 August 2026 for self-hosted environments in Claude Code, available to organisations on Team and Enterprise plans and switched off by default until an administrator turns it on. The mechanism is a single command, claude self-hosted-runner, that turns machines or containers an organisation already controls into the compute layer where Claude Code sessions actually execute — whether the session was started from the web, mobile, desktop, the terminal or a scheduled routine. Repository checkouts, build artifacts, secrets and anything a session creates or modifies stay on infrastructure the customer provisions, next to their existing network, toolchain and security controls, rather than running on Anthropic’s cloud.

This is a narrower announcement than it sounds, and that’s the point. It doesn’t change what Claude Code can do — it changes where the work happens, which is exactly the objection that has kept AI coding agents out of environments with strict data residency and compliance requirements. Regulated sectors, healthcare among them, have spent the last year watching AI-assisted development accelerate everywhere except inside their own walls, precisely because “send our codebase and patient-adjacent systems to a third party’s compute” was a non-starter regardless of how good the tool was.

So what

If your organisation has held off on AI-assisted development specifically because of where the code and data would need to sit, this beta is worth a proper look rather than a repeat of the same “not yet” answer. It’s early — public beta, opt-in, Team/Enterprise only — so treat it as a serious option to evaluate rather than a solved problem to roll out immediately. We build software for regulated sectors including the NHS and healthcare providers, and increasingly that means designing AI-assisted workflows that hold up to the same compliance and data-handling scrutiny as everything else in the stack. See our healthcare software development work, or get in touch if you’re weighing how AI coding tools fit into a compliance-constrained environment.

← All research

More research.

GitHub Copilot adds Supabase, Webflow, Wix and five more connectors in public preview

GitHub Copilot now connects to eight more services, including Supabase, Webflow, Wix and Mixpanel, through verified MCP connectors in public preview. They join more than 25 existing ones.

Read ›

Gemini 4 Neon is generally available in GitHub Copilot, aimed at lower-cost agentic coding

Gemini 4 Neon is now generally available in GitHub Copilot for agentic coding. Google says its quality is close to Gemini 4 Carbon at a lower token cost per task.

Read ›

Claude Code 2.1.328 brings auto mode to Bedrock, Vertex and Foundry, and narrows /code-review to bugs

Claude Code 2.1.328 extends auto mode to Claude Opus 4.6 and Sonnet 4.6 on Amazon Bedrock, Google Cloud and Microsoft Foundry. /code-review now reports only bugs and CLAUDE.md violations.

Read ›

GitHub Copilot Evals: repeatable tests for AI agents, now in public preview

GitHub Copilot Evals lets teams write repeatable tests for AI coding skills, custom agents and MCP tools. It checks both the result and the path the agent took, and is in public preview.

Read ›

GitHub Copilot can now drive your browser: browser use enters public preview

GitHub Copilot browser use lets an AI coding agent control Chrome to debug web apps and run browser tests. It is in public preview across the CLI, the Copilot app, VS Code and JetBrains.

Read ›

Claude Code 2.1.327 refuses some symbolic links and lists broken setup in /status

Claude Code 2.1.327 tightens file and package permissions again and adds a /status list of broken agent, skill and hook settings. A short read for teams using AI coding tools on a live product.

Read ›

Start a project

Building with AI? Let's talk.

Tell us what you are building. We'll give you an honest view of the right approach.

Start a projectSee our work