Claude Doesn’t Know Claude

The promise of AI, at least for me, is a system that matches my brain. No code. No hidden rules. Just language.

I use Claude three different ways. On three different systems at once: Chat, Code, Projects, Co-Work, Mobile. All of them called Claude.

That’s the test of the promise: can the system keep up with how I actually think?

After struggling with this all weekend, I realized the different Claudes don’t know what the others are doing. I finally asked Claude to build me instructions. Instructions so I could use all of them together.

The pieces don’t know each other exists. The system promised to keep up with my brain. Instead, it makes me remember all the pieces.

I had to ask Claude to help me work around Claude’s broken parts. Because I don’t know enough about AI to fix it myself. I didn’t even know there were different instructions for Claude Mobile until I started asking for help.

I have ADHD. I cannot remember what I did three minutes ago. Let alone which version of which spreadsheet we created last week.

The tool that was supposed to free me from that work is now requiring me to do more of it.

P.S. I wrote this field note on Mobile this morning. The instructions held. Don’t ask me if they survive Co-Work.