I started with ChatGPT. Then I discovered Claude.
Claude wrote better for me. Its phrasing felt more natural and closer to what I would say myself.
So I stayed with Claude for years.
Then Claude got worse for my work. Less reliable. And I kept running into usage limits.
After taking another look at the market, I switched to GPT-5.6 Sol without hesitation. My whole team came with me.
It worked extremely well.
Then Claude Opus 5.5 arrived.
The previous Opus versions had not convinced me. 5.5 is different.
Suddenly, both are good.
Today I use ChatGPT and Claude in parallel. Sometimes I choose based on the task. Sometimes I simply use whichever model I prefer that day.
And sometimes they work on the same project.
Then I appoint one as the lead AI. It handles the main task. The other AI evaluates the result and looks for errors, contradictions, or weak points.
Then I decide.
At the beginning of this year, switching like this would still have been cumbersome.
I would have had to transfer projects, rebuild the context, and explain how we work to the next AI.
Today my projects live in their own operating systems.
There, ChatGPT and Claude find the working rules, decisions, and current state. Both understand how the systems have evolved straight away.
So I no longer move large projects between platforms. I open the same system with whichever AI I want to use.
That is how quickly working with AI has changed.
Within a few months, choosing one platform went from a fundamental decision to a daily choice inside my workflow.
My projects are now anchored in my own system. ChatGPT and Claude take on different roles inside it.
And if another version becomes better tomorrow, I simply switch.
If you would like to learn more about how I work with these operating systems and how to build this structure for your own work, you can find it in my AI-First courses.
The 2026 course dates are already fully booked. Information about the programme and the 2027 dates is available on the Learning in the Sun course page.