This content originally appeared on DEV Community and was authored by Wojciech Lepczyński
Are you tired of AI assistants that send your every query to a massive cloud server, leaving you to wonder what happens to your data? The lack of privacy and control was a huge frustration for me, so I decided to do something about it. I built a completely local AI assistant that runs on my desk, is powered by Ollama from my old Laptop, and never, ever sends my information to the cloud. This is a story about taking back control of my personal data.
The ‘Why’ Behind the Build: Taking Back My Data
In a world where almost every smart device is connected to the internet and constantly transmitting data, the idea of a truly private assistant seemed like a dream. But it doesn’t have to be. I wanted a personal AI that was fast, secure, and most importantly, private. By building a self-contained system, I could guarantee that all my conversations and requests would stay within my own network. This project is a testament to the power of local-first technology and the freedom it offers.
How It Works: A Simple, Private Setup
The heart of the project is the CrowPanel Advance, a compact and versatile device with esp32 that serves as the assistant’s user interface. It’s connected via WiFi to a dedicated AI server that I set up on an old laptop, which runs the Ollama software. This is the magic part: Ollama allows me to run powerful large language models locally, so all the processing happens right there.
The CrowPanel simply sends a request to my private server, which processes it and returns a response. The entire conversation remains on my private network. It’s safe, it’s local, and it’s a brilliant example of what you can do with a little hardware and a lot of imagination. Oh, and its best feature? It can tell unique jokes whenever it’s bored, and it’s often bored! 
What’s Next? Stage 2 and Beyond
This is just the beginning. This first stage focused on building the core functionality and proving that a private, local AI assistant is not only possible but also practical and fun. For the next phase, I have big plans:
- A full graphical user interface (GUI): To make the assistant even more intuitive and user-friendly.
- Text-to-speech (TTS): To give the assistant a voice and make interactions more natural and hands-free.
- And more! The possibilities are endless.
If you’re curious about how I built this system and want to follow along as the project evolves, you have to watch the video.
This content originally appeared on DEV Community and was authored by Wojciech Lepczyński