AI is everywhere today, used for just about everything. A big downside of most providers is that you have to pay close attention to the data you share with them. Fortunately there's a relatively cheap and safe way to put AI to work for you without handing over sensitive data to third parties.
Using Meta's Ollama language models (open source), I set up an AI server that works like ChatGPT. But the possibilities are endless — you can combine models, fine-tune them, or build your own based on your own files. You can also connect a domain so the AI server is reachable from outside your home network. All of this runs on an ordinary Windows PC through Windows Subsystem for Linux (WSL), which behaves like a Linux server. If you also have a decent NVIDIA GPU in your laptop or desktop, it drastically improves performance.
In my view the biggest advantage is that you stay fully in control of what you feed into the AI, what information it has, and when you want to wipe that information.