I received an email from Atmoskop.cz. To the forwarded emails of former colleagues, so now I can finally add the one and true truth to their unsolicited feedback, muhehe 😈. And I realized how it will proceed with Our next contender for work.
What interests you about applicants? Internal integrity, personal maturity, professional competence?
For potential candidates, I am interested in their Claude.md files, the depth and structure of their context and the state of SKILLS. Of course, also how much they need a person to work.
Now you tell yourself one of three things.
1️⃣ Stupid joke! Ha, ha, ha! Ha! Hmm. Are you sure you know what AI can do right now? It is possible that you are doing a bunch of work needlessly.
2️⃣ Option two is that you say to yourself: “First of all, we recruit people and I guess they'll bring their AI agents! “
We haven't hired a human... hmm... for a year and a half? We don't plan to hire anyone new this year. Not because we don't want to prosecute more things, but because it doesn't actually make sense to us at all. We don't have a suitable job description. We change the workflow every quarter. To pick up a rookie would be an ordeal for both sides.
Of course, we're hiring agents. And actually, it bothers us a lot to work efficiently and for them to work properly.
3️⃣ Well and the third option is, that you take it completely seriously. And that's how I meant it.
For potential candidates, I am interested in their Claude.md files, the depth and structure of their context and the state of SKILLS. Of course, also how much they need a person to work.
For the last quarter, we are in the House of Řezč again changed working practices.
1️⃣ We all use Claude Code in terminal. We've moved from various desktop apps and other big language models to Claude Code and we're using it right in the command line. All including back-office.
2️⃣ We are trying to find a suitable directory structureso that Claude Code could work effectively in it. We all have a little different, mostly variations on the PARA system: Projects, Areas, Resources, Archive.
3️⃣ We were able to retell Claude what we were doing and how we do it. And then let yourself be helped. Because AI invents new workflows better than if you gradually fed tiny bits of your know-how into it and tried to automate or semi-automate current workflows.
How are you doing?
“It's now clear that using AI agents is itself a skill and the question is more if & how you teach people these skills versus debating whether AI tools make people more productive.”
— Dare Obasanjo
Reading for the weekend
Surely you have ever noticed in a discussion that someone accuses you of an argumentative foul. Maarten Boudry argues that making a pure argumentative foul is almost impossible.
