Fast sites. Isolated accounts. Your servers.
A complete hosting platform — web, mail, DNS, databases and backups — with the strongest per-account isolation of any control panel, a web stack built for speed rather than inherited from 2009, and mail that configures itself.
This is a preview of what is coming. It is being built in the open and is not finished; there is nothing to install and nothing to buy yet. What follows is what already runs.
containers · zones · certificates · mailboxes · snapshots · fleets
See what it does How it will be licensed
Isolation that is actually isolation
Other panels give each customer a unix account on a shared filesystem and call it separation. Here every account is its own container: its own dataset, its own process space, its own runtime, reached over its own socket. There is no shared document root and no path from one tenant’s code to another’s files.
Ceilings are real, not advisory. CPU, memory and process limits per plan, and outbound bandwidth shaped on the host side of the interface — where nothing inside the container can opt out of it. Even the per-account object cache is a private instance on a private socket, so there is no shared keyspace to leak across.
One tenant’s runaway process is not another tenant’s outage. That is the foundation the rest of the product is built on, and it is why the rest of it can be simple.
Built for speed, measured rather than claimed
A web server we compile ourselves, with HTTP/3 and QUIC, Brotli beside gzip, and a full-page cache that can purge a single URL — which stock nginx cannot do at all, and which is what makes a page cache safe to leave switched on.
Per-account object caching, and OPcache with JIT and optional preloading. When something is slow, the panel names the script rather than shrugging: real-user timings per site, slow-request logs, and per-query visibility.
Six runtimes, all current: PHP 8.2, 8.3 and 8.4 side by side and selectable per domain, Node.js 22, Python 3.11, Ruby 3.1, .NET 8 and Java 21 — each behind the same socket contract, so a site changes stack without changing anything around it.
Mail that sets itself up
Your customer types their address and their password into Thunderbird, Apple Mail or Outlook, and it works. No server names, no port numbers, no SSL/TLS dropdown, no support ticket. The panel answers the discovery protocols every mail client already speaks — including Outlook’s, which asks a different question in a different place than everyone else’s.
That one feature removes what is, on most hosts, the single largest category of support contact. Everything underneath it is handled the same way: the records that decide whether your mail is trusted are generated, published and watched rather than left as a checklist somebody forgot.
A sandbox for every site, for 8 KB
Cloning a 64 MB site into a staging copy allocates 8 KB. That is a measured number, not an estimate, and it holds because a staging site here is a copy-on-write snapshot rather than a copy — it starts costing disk only for the bytes that actually change.
So staging stops being something you ration. Every customer can have one on every site, break it, throw it away and make another, and your disk barely notices.
Backups that are proven, not promised
Every host claims backups. Ask one when a restore was last actually tested and the answer is usually silence, because testing a restore normally means finding a spare machine.
This panel restores one every week and checks the site still renders. It picks the account whose backup has gone longest unproven, restores it into a throwaway container, puts a throwaway web server in front of it, and asks the restored site for a page — without touching the live one. A backup missing the files that make a site work passes a checksum and fails this.
Accounts leave the node as encrypted snapshots, and the key that reads them is sealed to a recovery key you generate in your own browser. We never see it and cannot be compelled to produce it. Losing a server should be an afternoon of rebuilding, not a phone call you dread — and that is a claim we test on your behalf rather than ask you to trust.
The exits are never locked
Backups, restore, recovery, certificates and a complete account export will be on every plan, including the free one — no exceptions, and a test in the source that fails if anyone tries to change it.
A lapsed subscription drops a node to the free tier. It does not go read-only, and it has never been able to stop a website, a mailbox or a database: the licence check does not reach any of them, by construction. Commercial disagreements are between you and us, and your customers should never be able to tell.