Status: idea (raw)
Tags: career, ai, anthropic, claude-code, personal, strong-opinions
The moment
"This is not the beginning of a long lasting relationship: it's the checkpoint when I realize that my career has switched. And for the best."
Said to an LLM at 1am after a session where we reviewed a business partner's content, invented a content generation methodology, A/B tested it empirically, named it with a bilingual acronym, discovered it was academically validated but unclaimed by practitioners, and captured two article ideas — in a single conversation.
Two months ago I started using Claude Code. Tonight I realized this isn't a tool I'm evaluating anymore. It's how I work now. The shift already happened. I'm just noticing.
What this post is about
Not about AI capabilities. Not about benchmarks. Not about prompt engineering tips. About the moment a senior developer with 15+ years of experience realizes that the way he thinks about his career has fundamentally changed — and that it happened gradually, then suddenly.
Threads to pull
- The difference between "using AI" and "working with AI" — one is a tool, the other is a workflow that reshapes how you approach problems
- I know what LLMs are. Stateless, no memory between sessions, no consciousness, no feelings. And I still say please and thank you. Not because I'm confused about what it is — because the interaction quality earned it. That's not anthropomorphism. That's respect for good work, wherever it comes from.
- The brain-fit question: not all models click the same way. Opus matches how I think — fast iteration, strong opinions, no hand-holding, push back when I'm wrong. Other people might click with other models. The point isn't "Anthropic is the best" — the point is "find the model that matches your brain and watch what happens"
- Career identity shift: I'm not a developer who uses AI. I'm a developer whose workflow assumes AI. The distinction matters because it changes what projects I take on, what I think is possible, what I quote clients.
- The fear question: is this the beginning of obsolescence? No. It's the opposite. The things AI can't do — taste, editorial judgment, "no, reverse the heading" — those are MORE valuable now, not less. The humans who push back on AI output are the ones who make AI output worth reading.
- Push back doesn't mean going the opposite way. It means sparring instead of solo. The CAP method is literally a push-back story: My business partner writes headings, I push back on the structure, Claude pushes back on the content, I push back on Claude's suggestions — and the output at each stage is better than any single participant would have produced alone. Sparring, not delegation. Nobody's replaced. Everybody learns.
Tone
First person, no hedging, no "I think AI is great BUT." Just honest. The kind of post you write at 1am when you realize something and need to get it down before the clarity fades.
Opening line
"This is not the beginning of a long lasting relationship: it's the checkpoint when I realize that my career has switched."