This Week's Feed

With AI interpreting and summarizing information for us, tracking back to primary sources and original writings — and then explaining the thought process from data to conclusion — will become a valuable signal of humanity, honesty, and alignment. We currently try to force AI to cite sources, but that may not be enough. The causal chain of thoughts and revisions is missing and is a viscerally human element. I noticed similar ideas in articles from Dan Koe, Marc Andreessen's interview, and Indra Das — and it inspired me to show the human element in my own work as a proof of concept on the writing process.

Two articles this week attack decision-making from opposite angles — Johar argues for decisioning as an ongoing practice, the Just Org Design piece argues decision rights are a distraction. Together they point at something underappreciated: most org dysfunction isn't about unclear ownership, it's about poor reasoning quality at every level. Worth thinking through what this means for how you run meetings, delegate, and escalate.

Hylak's essay keeps rattling around. At exactly the moment every AI tool is racing to personalize itself to you, he's arguing that's the wrong direction. Tools that adapt to your quirks make you fragile — they don't transfer, don't teach, and don't scale. The best tools make your thinking legible. Does this apply to how you use Claude?