Claudin' Around

I rebuild this site every time I decide I feel like writing again, but at least this time it's a good excuse to play around with Claude. All told, it's impressive: after a few brainstorming conversations over the last week and and two hours of iterative noodling I have a working build. Better still, I feel halfway decent about it and it stubbed out a few wild ideas I'd like to explore. I'm certainly not pushing the envelope or anything—still getting my feet wet—so I'm mostly letting it jam in the background while I work on other things. But I'd say the "wrap myself around the axle with overthinking things until self-loathing gets the better of me" problem feels addressed which is awfully nice. Stuff I want to get better at:

• Actually thinking rather than just delegating. Most of what I've asked so far feels very much like the kind of stuff I could've done on my own with enough time. (e.g. Picking a framework, scaffolding the project, migrating content.)
• Cross-referencing work and using adversarial approaches. Seems like I'm seeing a fair number of folks I follow are moving toward using a few agents/processes to work, check work, and counter balance one another. (Overkill for rebuilding a blog, but I've been revisiting some of my ideas for larger utility applications for experimentation.)

To be honest, I'm most surprised by how much (and how often) I feel compelled to be polite to Claude. It just feels... wrong to speak in natural language, making requests, without some sort of gratitude? It's easy to compare agaisnt older folks saying please and thank you in their Google queries, but this is the first time I've felt this way despite being exposed to AI in search, Copilot at work, etc. For the time being I'm treating it as amusing—worst case is that I've been deferential to some sort of consciousness and contributed to midwestern stereotypes.