Mike MurphyAI Handyman
Subscribe
Back to tutorials

Run a Private AI Chatbot with Ollama and Open WebUI

May 22, 2026 ·

Set up a private, self-hosted AI chat interface on your Hostinger VPS using Tailscale, Ollama, and Open WebUI.

Description

In this tutorial, you will learn how to set up a private, self-hosted AI chat interface on your Hostinger VPS using Tailscale, Ollama, and Open WebUI. This is Part 2 of the Tailscale series. Once it is set up, you will have a ChatGPT-style chat interface running 24/7 on your own server, accessible from any device on your Tailscale network, completely private, secure, and free (beyond your existing VPS subscription).

🧠 What You Will Learn

  • Open WebUI… deploy it as a Docker container on your Hostinger VPS
  • Ollama… install it as the AI engine powering your chat interface
  • Docker networks: confirm both containers are on the same network
  • Claude Code: use it to deploy Docker containers on your VPS via SS
  • YAML config: set ports correctly and keep services internal only
  • Tailscale MagicDNS: connect to Open WebUI from any device on your network
  • Model management: pull new Ollama models and switch between them in Open WebUI

🔗 Links

🌐 Hostinger VPS (Affiliate Link):
https://www.hostg.xyz/SHIDN

Tailscale:
https://tailscale.com

Open WebUI https://openwebui.com

Ollama: https://ollama.com

⏱️ Timestamps

00:00 Intro: About The Tutorial
01:11 Tools Used
02:05 Tools Overview
02:54 VPS: Docker Catalog
04:10 Claude Code SSH
05:20 VPS Skills
06:57 Confirm Ollama & Open WebUI
07:43 Docker Network
09:45 Open WebUI Docker Compose
11:26 Check Tailscale Network
12:22 Tailscale: Magic DNS
12:59 Connect to Open WebUI
14:31 Connect to Open WebUI in Browser
15:08 Open WebUI Overview
16:41 Download Ollama Models

What We’re Building

  • Ollama — runs local AI models (like Gemma, Qwen, Kimi, etc) on your VPS
  • Open WebUI — a polished ChatGPT-style browser interface that talks to Ollama
  • Tailscale — a free private network that connects all your devices securely without opening ports to the public internet
  • Both Ollama and Open WebUI are running in Docker containers on your VPS
  • This tutorial focuses on the wiring and the connection — not spinning up the containers from scratch

The Big Picture (Why This Setup is Worth Understanding)

I know it sounds like a lot of moving parts — but think of it like this:

“Ollama is the engine. Open WebUI is the dashboard. Tailscale is the private road that lets you drive from anywhere.”

Layer Tool Job
AI model runner Ollama (Docker container) Runs the actual AI models on your VPS
Chat interface Open WebUI (Docker container) Gives you a browser-based UI to talk to Ollama
Private access Tailscale Securely connects your devices to your VPS — no public ports needed

Why not just expose Open WebUI publicly? You could — but then anyone could potentially reach it. Tailscale keeps it completely private to only your devices, with zero extra firewall configuration.

Why run this on a VPS at all? Your local machine doesn’t have to stay on. The VPS runs 24/7. Pull up your phone in the airport, open a browser, and your AI chat interface is just there.

Prerequisites

  • A VPS running Ubuntu (Hostinger or similar)
  • Ollama already running in a Docker container on that VPS (link in description)
  • Open WebUI already running in a Docker container on that VPS (link in description)
  • Both containers are on the same Docker network
  • SSH access to your VPS (Termius or terminal)
  • A Tailscale account — free tier is all you need (tailscale.com)
  • A device to test with (MacBook, iPhone, etc.)

⚠️ “I’m not going to walk through setting up Ollama or Open WebUI in Docker in this tutorial — I’ll link those in the description. If you’ve got both containers running, you’re in the right place.”

Connect to Your VPS

SSH into your VPS or open the terminal in your hosting dashboard.

Step 1: Confirm Both Containers Are Running and On the Same Docker Network

docker ps

You’re looking for two containers — one for Ollama, one for Open WebUI — both listed as Up.

docker network ls

Find the shared network name (something like ai-network or whatever you named it).

docker network inspect <your-network-name>
  • Look for both container names listed under Containers
  • If they’re both there — you’re good

⚠️ “If they’re on different networks, Open WebUI won’t be able to reach Ollama by container name. That’s the most common issue at this stage.”

Step 2: Confirm Open WebUI Can Reach Ollama

Inside Open WebUI, Ollama is referenced by its container name, not localhost. This is how Docker networking works — containers talk to each other by name on a shared network.

The connection string should look like:

http://ollama:11434

Where ollama is the container name for your Ollama container.

📝 You set this in Open WebUI’s environment variable — typically OLLAMA_BASE_URL=http://ollama:11434 in your Docker Compose or run command.

  • Open a browser on your local machine and navigate to your VPS IP on whatever port Open WebUI is running (e.g. http://YOUR_VPS_IP:3000)
  • Log into Open WebUI
  • Go to Settings → Connections and confirm Ollama URL is pointing to http://ollama:11434
  • You should see your models listed — that’s your green light

⚠️ “If you see a connection error here, it almost always means the containers aren’t on the same Docker network, or the Ollama container name doesn’t match what you typed.”

Step 3: Install Tailscale on Your VPS

Tailscale installs on your VPS like any other package. One command.

curl -fsSL https://tailscale.com/install.sh | sh

This pulls the official Tailscale installer and sets it up on your server.

Step 4: Authenticate Your VPS to Your Tailscale Network

sudo tailscale up
  • This outputs a URL — copy it and open it in your browser
  • Log in with your Tailscale account (Google, GitHub, etc.)
  • Your VPS will show up as a new device in your Tailscale dashboard

📝 You only do this once. After this, Tailscale starts automatically on reboot.

Step 5: Get Your VPS Tailscale IP

tailscale ip -4

This gives you something like 100.x.x.x — your VPS’s private Tailscale IP address. Write this down.

📝 This is a private IP that only your Tailscale devices can see. It does not expose your server to the public internet.

Step 6: Install Tailscale on Your Other Devices

On your MacBook: download from tailscale.com/download and sign into the same account.

On your iPhone/Android: install from the App Store or Play Store, sign in with the same account.

Once installed and logged in — your devices are on the same private network. That’s it.

Device Install Method
MacBook Tailscale app — tailscale.com/download
iPhone App Store → Tailscale
Android Play Store → Tailscale
Another VPS/server Same curl installer command

Step 7: Access Open WebUI From Any Device

On any device connected to Tailscale, open a browser and go to:

http://100.x.x.x:3000

Replace 100.x.x.x with the Tailscale IP from Step 5. Replace 3000 with whatever port your Open WebUI container is mapped to.

  • You should see your Open WebUI login screen
  • Log in — pick a model — start chatting
  • You’re talking to Ollama on your VPS, through your private Tailscale network, from any device in the world

“And that’s pretty much it. Private AI chat, on your own server, from anywhere. Pretty cool.”

Step 8: Verify It All Works End-to-End

Quick sanity checklist before you call this done:

  • Open WebUI loads in the browser via Tailscale IP
  • Models are listed in the model dropdown (pulled by Ollama)
  • You can send a message and get a response
  • Test it from a second device (phone or laptop) — confirm both work
  • Disconnect from your home WiFi on your phone, use cellular — it should still work via Tailscale

⚠️ “If it works on your laptop but not your phone, just double-check Tailscale is running on the phone and you’re signed into the same account.”

Core Commands Reference

Docker checks

docker ps
docker network ls
docker network inspect <network-name>

Tailscale on VPS

curl -fsSL https://tailscale.com/install.sh | sh
sudo tailscale up
tailscale ip -4
tailscale status

Confirm Tailscale is running after reboot

sudo systemctl status tailscaled

Complete Setup Summary (VPS Side)

# 1. Confirm containers are running
docker ps

# 2. Confirm they're on the same Docker network
docker network inspect <your-network-name>

# 3. Install Tailscale
curl -fsSL https://tailscale.com/install.sh | sh

# 4. Connect VPS to your Tailscale network
sudo tailscale up

# 5. Get your Tailscale IP
tailscale ip -4

# 6. Access Open WebUI from any Tailscale device
# http://100.x.x.x:3000

This is the clean wiring setup — Docker networking handles the Ollama connection, Tailscale handles the private remote access.

Wrap Up

You just turned your VPS into a private AI chat server you can reach from anywhere in the world — on your phone, MacBook, tablet, whatever — without opening a single public port. Ollama runs the models, Open WebUI gives you the interface, and Tailscale creates the private road between all your devices.

  • Ollama and Open WebUI talk to each other inside Docker via container name ✓
  • Tailscale creates a private network — no public exposure needed ✓
  • Any device with Tailscale installed can reach Open WebUI from anywhere ✓
  • Free tier covers everything shown in this tutorial ✓

My name is Mike Murphy, your AI Handyman — cheers!

Questions, Comments, Feedback? hello@mikemurphy.co

My Gear List: https://mikemurphy.co/resources

My Amazon Store: https://www.amazon.com/shop/mikemurphyco

My YouTube Channel: https://youtube.com/mikemurphyco

AI Unplugged · weekly

Newsletter for AI Builders

One practical email with the news, tools, terms, and tutorials worth remembering. Beehiiv sends it; Astro owns the archive.

No spam. Reply-friendly. Easy to leave.