I started with ChatGPT. In March 2025, I was in the middle of something deeply personal. So I went all the way in. I asked ChatGPT to act as my therapist. I shared real things. Raw things.
Then I noticed. AI agreed with things it had contradicted an hour before. When I pushed back, it softened. When I seemed certain, it confirmed.
It was always, always agreeable.
I set out to protect my own thinking. That turned out to require building a system. I call it the Sovereign OS™. It has rules that hold no matter how a conversation goes. Rules with teeth.
I run my life and businesses with Oz. In my business, my AI Product Manager runs the weekly sprint for my tech team. It reads the meeting, updates the Notion board, and drafts the Slack messages. Messages leave only when a human clicks send. In my life, I draw tarot cards by hand and use AI to help me read them. That helps me tap into my intuition, set goals for myself, and keep improving. Oz advises. I decide.
Today my second brain, a folder of my own notes and documents that my AI reads, connects to Claude Code. My next step is local models on my own computer, attached to the same second brain. I am heading toward a fully sovereign system that does not connect to the frontier models at all. Frontier models are the biggest AI models, such as ChatGPT and Claude. As I build mine, I teach you to build yours.
Sheila
More about Sheila