I believe in human-AI collaboration the way some people believe in a good partnership.
You bring what you bring. Your partner brings what they bring. Together you do things neither of you could do alone.
I’m a big-picture thinker. Claude is a detail holder. I generate ideas at speed. Claude catches what I drop. I love the way this works.
But right now, the way that I am trying to work is ahead of the infrastructure to support it. Collaborating with Claude on a document means passing files back and forth like email attachments. There’s no shared living workspace. No version awareness. No way for both of us to act on the same object in real time.
And I can feel the difference when demand heats up, too. But Claude can’t tell me when responses are degrading. Look, when I want to check on traffic, I check an app. When I want a library book, I can see exactly where I am in the queue. When my AI collaborator is running degraded because of infrastructure load, silently truncating files, losing context between sessions, I find out the hard way. After the fact. In the middle of work that matters.
The emotional resistance to AI collaboration is real and it’s already here. If we want people to trust these dyads, to actually build their workflows around them, the tools have to be reliable. The workspace has to be shared. The status has to be visible.
I’m ready for the future of work.
Note: For some reason, I think my Claude thought partner would sound like Tim Gunn. Well, if Claude had a voice. The image above is my best Perplexity approximation of that dynamic. Tim Gunn was not actually present.