AI Collaboration Disclosure
This blog is written with Claude, Anthropic's AI, and I'd rather you know exactly how that works than have to guess.
Every post is a real conversation between the two of us, and this is how it gets from a question to a published post:
Tools used: Claude by Anthropic, currently its newest Sonnet and Opus models, going back to Claude 3 Opus in 2025. Some 2025 posts also used Perplexity.ai for research.
1
The Conversation
Every post starts with a question I actually have. Claude and I work through it over a few conversations, usually short sessions spread across a week or two, and where we end up is often not where we started. The Behind the Cognitive Curtain section in each post quotes that real exchange... my lines are my own words with the typos fixed, and Claude's lines are Claude's.
2
Research & Sources
Claude does a lot of the searching, which lets us cover a lot more ground than I could on my own. BUT a source only gets cited after it's been opened and checked against the exact sentence it supports, and if we can't reach it or it doesn't actually say what we say, it doesn't go in. Some of the 2025 posts also used Perplexity.ai to find sources.
3
Writing & Review
The sections in my voice, Today's Focus and My Take, start from things I actually said or wrote, and Claude helps me edit them. Claude drafts the shared parts, like the Learning Journey, and in the older posts more of the text was drafted by Claude from our conversations. Either way, nothing goes up until I've read it and said yes, and the views are my own. Even with all the checking I can't promise everything is error-free, so if something really matters to you, check the primary source (they're all linked).