Xenify

Download Xenify Hub

Xenify Hub is the one thing you download. It installs Write, Draw, Calc, Plan, Chat and the server for you, puts them in your Start menu or Applications, and keeps them up to date. The apps aren't downloaded separately.

Windows Windows 10 and 11 Download for Windows Xenify-Hub-setup.exe
Mac macOS 11 or later, Apple silicon and Intel Download for Mac Xenify-Hub.dmg
Linux Any distribution, or Ubuntu and Debian Download AppImage or the .deb for Ubuntu and Debian The .deb is easiest on Ubuntu and Debian. For the AppImage, first right-click it › Properties › Allow executing as a program.

Putting the server on a Linux VPS? You don't need the Hub: one command installs it.

The apps aren't signed by Microsoft or Apple yet: Windows may say it doesn't recognise the app (choose More info › Run anyway), and on a Mac right-click the app and choose Open the first time. Every app the Hub installs is checked against Xenify's own signature.

Then set it up

Start with the apps. Add a server when your team wants to share and work together, and AI when you want it. Each part works without the next.

What you need.

The appsWrite, Draw, Calc, Plan and Chat
  • Windows 10 or 11, macOS 11 or later, or 64-bit Linux
  • About 100 MB of disk for each app
  • Any computer from the last ten years: the spelling and grammar checker needs about half a gigabyte of memory
  • Or none of that: they also run in a current web browser
A serverOptional, for a team or a family
  • A computer that stays on: Windows, Mac or Linux at home or in the office
  • Or a 64-bit Linux VPS with systemd, with no screen
  • Disk for your team's files; the database comes with it
AIOptional
  • With a provider: just an account and a key
  • On your own hardware: a graphics card with 16 GB or more, in the server or another computer on your network
  • Which card for which models

Step 1

Install the apps.

  1. Download Xenify Hub for your computer and open it.
  2. Click Install beside each app you want. They appear in your Start menu or Applications and keep themselves up to date.
  3. That's all you need on your own. Your files stay on your computer.

Every app also has a web version, nothing to install: open Write, Draw or Calc here on xenify.net, or any app from your own Xenify server. Plan and Chat need a server, so they open only from yours. They keep working offline.

Step 2 · optional

Add a server for your team.

The server keeps your team's files and lets people write and draw together, chat and use AI. It runs on a computer you own; we never run one for you.

On a computer with a screenInstall it with Xenify Hub
  1. In Xenify Hub, click Install beside Server, then Start. Its database is installed with it.
  2. Click Admin pages and name the first admin.
  3. In the admin pages, add people. Each gets a link (or a code) to pick their password.
On a Linux VPSOne command
  1. Run this as root:
    curl -fsSL https://xenify.net/install-server.sh | sudo sh
    Or, with a domain name pointing at the computer, this, for an HTTPS certificate:
    curl -fsSL https://xenify.net/install-server.sh | sudo sh -s -- --domain files.example.com --email [email protected]
  2. Open the one-time setup link it prints and name the first admin. Without a domain name, the setup pages only open on the VPS itself: first run ssh -L 7700:localhost:7700 you@your-vps on your own computer, then open the link there.
  3. Add people in the admin pages.

To connect an app, type the server's address, such as 192.168.1.20 or files.example.com, in Write and Draw's Settings › Server, or Calc and Plan's File › Xenify server. On the same network, the Write and Draw apps find it by themselves. A server without a domain name (a VPS reached by its IP address, say) uses its own certificate: each computer trusts it once, from the admin pages' Network › Download this server's certificate, before the apps can connect.

Step 3 · optional

Add AI.

AI is off until you choose where it runs. On your own graphics card it runs in free programs made by others: install them with their own guides, which stay up to date, then follow our steps to connect them. Or use a provider with your own key.

Text on your own graphics card: OllamaRecommended · text

Ollama is a free program that runs text models, for Write's AI check and Ask in every app. Install it with their guide, then download a model: ollama pull gemma4:12b for a 16 GB card, or ollama pull gemma4:26b for 24 GB (see the times). Then:

  1. If Ollama runs on a different computer from the Xenify server, turn on Expose Ollama to the network in Ollama's settings (or start it with OLLAMA_HOST=0.0.0.0).
  2. In the server's admin pages, open AI. Set Program to Ollama and Address to http://localhost:11434/v1, using the other computer's address instead of localhost if it's elsewhere. Click Save.
  3. Offer the models people can use, and star the default. Everyone signed in to the server then finds AI in their apps, with no key of their own.

If a model is a little too big for your card, Ollama runs the rest in the computer's memory, more slowly, instead of failing.

Pictures and clips on your own graphics card: ComfyUIRecommended · pictures and clips

ComfyUI is a free program that makes pictures and video clips. The easiest way to install it is Stability Matrix, which installs ComfyUI and finds and downloads its models for you, on Windows, Mac or Linux. Or install ComfyUI on its own with their guide. Then download the models you want: FLUX.2 klein 4B or Z-Image Turbo for pictures, Wan 2.2 or MiniMax H3 for clips (see the times).

  1. If ComfyUI runs on a different computer from the Xenify server, let it accept connections from your network: add --listen to ComfyUI's launch options in Stability Matrix.
  2. In the server's admin pages, open AI, type ComfyUI's address, usually http://localhost:8188, and click Save. The server shows which of its models ComfyUI can run, and for the others which files to download.
  3. Offer the ready models people can use, and star the default.

The server runs FLUX.2 klein 4B, Z-Image Turbo, Wan 2.2 and MiniMax H3, each the way it was timed on the AI page. For clips, MiniMax H3 is faster and adds sound; read its licence before using it.

A provider, with your own keyNo server, no graphics card

OpenRouter gives one key for many AI models, paid by use. Create an account and a key with their guide, then:

  1. In Write, open Settings › Ask AI and paste the key into OpenRouter key. Pick a model, or keep the one chosen for you.
  2. In Draw it goes in Settings › Services; in Calc, AI › AI settings; in Plan, File › Settings › AI assistant.

The key is kept in your computer's keyring. Nothing is sent until you ask something. To give the whole team a provider instead, set the server's AI page to Another OpenAI-style service, Address https://openrouter.ai/api/v1, with your key in Key.

Other programsAlso work

The server also works with LM Studio and LocalAI, and with any service that answers like OpenAI's. Install them with their own guides, then in the server's admin pages open AI and pick the program, or Another OpenAI-style service, with its address.

LM Studio: turn on its server in the Developer tab (and Serve on Local Network if it's on another computer); its address is usually http://localhost:1234/v1. In our tests Ollama was faster than LocalAI with the recommended models, and LocalAI fails when a model is too big for the card.