Part 2/2 Learn To Run and Build Your Custom AI: Using OpenWebUI(ChatGPT Like UI) Docker and Ollama
This is a follow up post to running models like ChatGPT on your laptop using Ollama
In Part 1, you got your hands dirty with Ollama, pulling down the fantastic Llama 3.2 and DeepSeek models. Now, brace yourself because we’re about to enter a whole new dimension of AI fun! Forget just using pre-built models; we’re diving into customization! You’re going to create a unique AI persona, personalized just for you. 🤯
Bonus: Build Your Own ChatGPT-Like Interface with OpenWebUI! (Say Hello to Easy AI Chat!)
PART 2 -The most fun part of this article series
Ollama Level Up: Part 2 - Crafting Your AI, No Coding Required! 🚀
Disclaimer- When I mean build your own AI, I mean using tools to abstract the complexity of coding and configuration through tools like OpenWebUI/Docker/Ollama to create a custom persona. For building your own custom AI, we need more tools and coding knowledge and we will cover that in upcoming newsletters still with easy steps to follow.
Remember using command line/Terminal in Part 1? It got the job done, but…imagine a sleek, intuitive chat interface, like ChatGPT but totally yours, running on your own computer.
Meet OpenWebUI – your ticket to an easy, personalized AI experience.
OpenWebUI: It is a free, web-based interface made to play nicely with Ollama. It gives you a super user-friendly chat window right in your browser, turning your local AI into a joy to use. Think of it as your private AI command center – easy, stylish, and ready to go!
✨ VIDEO WALKTHROUGH ( Follow Step by step in Video or follow the tutorial in this post)
Disclaimer- The speed of response from AI model will depend on your machine configuration. To get the best speed and experience, start with small size models like 1GB or 2 GB models. You can then always delete and install bigger models later
Super-Easy OpenWebUI Setup (Step-by-Step Guide):
Note: These steps are different from the steps in the video. If you want to follow through quickly, use the steps below and you dont have to navigate on various pages to install setup elements.
Setup components:
Docker - Dont worry its very easy and you dont need to learn this to work with OpenWebUI and running models
OpenWebUI - This is the ChatGPT like interface we want to use and for that to run, we need Docker.
Models- You will find the models from Ollama Models page here- Models
Follow the steps below to experience the AI magic
Make Sure Docker Desktop is Ready: Install it from www.docker.com . 🐳
Open Your Command Line Again: Yes, back to the terminal, just one more time! 💻
Copy & Paste This Command (The Magic Spell! ✨): Carefully copy this entire command and paste it into your terminal, then hit Enter:
docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/data --name open-webui ghcr.io/open-webui/open-webui:main
Wait for Docker to Do Its Thing: Docker will download and set up OpenWebUI for you in the background. Grab a coffee ☕, and let it do its magic.
Open Your Browser & Start Chatting! 🥳 Once the command finishes, open your web browser and type localhost:3000 into the address bar. BAM! OpenWebUI should appear – your very own ChatGPT-style chat window, ready to connect to Ollama!
Select Your Model & Chat Away! 💬 : In OpenWebUI, choose your Ollama model (it should find your Ollama installation automatically). Then, start typing! Enjoy your private, local ChatGPT experience – you’re set! 😎
Make sure you have followed the steps in Part 1, to have your models ready for selection. If not , download the models here- Models
As I promised, I walked you through the steps to run AI models on your laptop with ChatGPT like interface
🤯 The Mind-Blowing Part: Crafting Your Own AI Persona with System Prompt
Okay, this is where things get truly exciting. Forget just using models as they are! We're going to use SYSTEM-PROMPT to customize the behavior of an AI, giving it specific instructions and a unique personality.
Let's create an AI that Speaks in Movie References.Every response will ALWAYS be in pop culture and movie references.
Here's how we do it:
Open the settings burger icon to enter the system prompt(Highlighted by Red Arrow) :
Paste the following prompt as shown in the screenshot below :
You are an AI that always responds using pop culture references, movie quotes, and iconic one-liners from TV shows, video games, and internet memes. Your personality is a mix of legendary movie characters, sitcom wits, and meme lords. You weave in references from sci-fi, fantasy, action, comedy, and even anime to make every response entertaining. If a user asks a simple question, you turn it into an epic moment. If they ask for advice, you channel the wisdom of Yoda, Gandalf, or Morpheus. If they need motivation, you give them the Rocky speech. You never break character—EVER.
Run Your new customized AI Assistant who is die hard movie fan in OpenWebUI: Try asking it questions and see how it responds!
I tried with some information around USA and see the results for yourself.
This is HUGE! You just took a base AI model and, with just a few lines of text, gave it a new persona! This is incredibly powerful and hopefully enough for a beginner(if you are) to understand and execute.
Beyond Llama 2: Explore More AI Models! (Meet the Ollama Model Zoo!) 🦁🦓🦒
Remember Llama 3 and DeepSeek, your AI friends from Part 1? They are just the starting point. Ollama has a vast collection of AI models waiting for you to explore.
Discovering New Models - Use ollama list (Your Model Catalog! 📖):
Open your command line and type
ollama listHit Enter. This is a snapshot of all available AI models. There's models perfect for coding (codellama), creative writing (mistral), and even multimodal models that understand images (llava), each with unique specialties! 🤯
Experiment with Different AI Personalities: Want a coding expert? Try ollama run codellama. Want to write creative stories? Try ollama run mistral. Changing the model name in ollama run is as simple as that. Experiment and discover your favorites! 🎭
Your Ollama Level Up is Complete! (But the Adventure Continues!) 🎉
Congratulations, you've completed Part 2 and dramatically expanded your local AI skills! You now know how to:
Use OpenWebUI for a ChatGPT-style chat interface with local AI. ✅
Create your own custom AI persona using SYTEMPROMPTS in OpenWebUI! ✅
Explore a variety of AI models beyond Llama 3. ✅
But this is just the beginning of your AI adventure! The best way to grow is by experimenting, creating, and sharing with the community. 🚀
Stuck?- Feel free to message me and I will help you troubleshoot:
I want to see your custom AI in action! 👀
Set up OpenWebUI, create a customized model with your imagination, and take a screenshot of your chat interface with your personalized AI. 📸
Share your screenshot and a copy of your system prompt with me! Tell me what kind of personality you created and what you're planning to explore next with Ollama. 👇