self-hosted

A corporate VPN on your own servers, with a console that shows everything

Guard Tun runs on servers your company owns or rents. No third-party VPN provider sits between your code and the AI service, and the administrator sees the whole installation in one place.

your servers

Why run it yourself

A consumer VPN sends your developers’ traffic through a company you have to trust, out of an address that thousands of strangers share. Self-hosting removes both problems. The tunnel ends on a server that belongs to you, and the address the AI services see is the address of that server.

  • The egress is yours. Claude Code, Cursor and Codex reach their services from one address that only your company uses.
  • No VPN company in the middle. Your developers’ traffic goes from the device to your node and from there to the AI service.
  • You choose where the nodes stand. Pick the region, the hosting provider or your own hardware. The installation does not care which.
operator console

What the administrator sees

The console opens in a browser and runs on your server. It answers the questions an administrator gets asked without a terminal session.

  • Nodes, with their reachability and load.
  • Devices, with their status, client version and last connection.
  • Usage per user and per node, for today, for 7 days and for 30 days.
  • The audit log: every change made by an administrator.
your data

What stays on your side

The console and its data live on your server. Profiles and keys are issued there. The installation runs on your servers, so the questions of who can reach it, where it is backed up and how long records are kept are decided by your own policies, not by ours.

This is also the shorter conversation with your security and legal teams. There is no new processor to review and no vendor contract for the traffic itself: the servers are the ones you already know how to assess, and the installation is one more service running on them.

For the people who run it, that means one place to look. Nodes, devices, usage and the audit log are in the same console, and the audit log records what administrators did.

a pilot

What you need to start

  • A Linux server for the console and a node. For a pilot they can be the same machine.
  • A domain name for the console.
  • An administrator account.

Each developer installs a client and opens a profile link the administrator sends. That is the whole setup on their side. To plan a pilot on one of your servers, send the request form below.

the team

Who it is for

It suits an engineering lead who wants the whole team on one stable route to the AI tools, and a security person who wants to know exactly where the traffic goes. Both get the same answer: to your node, on your server, and out from your address.

faq

Questions

Do you host anything for us?

No. The console and the nodes run on servers you own or rent. Nothing about your installation is hosted by us.

Can the console and a node run on one server?

For a pilot, yes. One Linux server is enough to start with a few developers. As the team grows, it is sensible to give the console and the nodes their own servers.

What does the administrator see about a developer’s device?

Its status, the client version, the last connection and the traffic totals. The console shows how a device is doing, not the content of its traffic.

Who can open the console?

Administrators only. SSO and MFA for administrators are coming soon.

request

Request a demo or a pilot

Tell us about your team, and we will come back with a time for a call and a plan for a pilot.