On natural language as the next level of abstraction: the optimistic comments in this thread are interesting, i.e. because of the productivity and intelligence of agents, a high-level understanding of the system is “good enough” and it’s less important to know your way around the underlying code now. My own approach to agentic development in my personal projects has largely been hands-on and supervised (i.e. I still review generated code) but even I’ll admit that with how good Fable and Opus 5.5 have been recently, my smaller, single-purpose apps have gradually gone the way of blind trust and “make no mistakes”. Obviously, I don’t apply this to code meant for production, but discussions like these make me wonder how far away we are from blind trust in agents becoming the norm. Talking strictly about writing code (i.e. not requirements gathering or architecture etc.), will a human in the loop always be necessary? https://www.reddit.com/r/AgentsOfAI/s/ZGlPLS1glz
Quick Notes
Our Common Time • Kyle Sarte • 2026
5 quick notes
October 2026 Notes:01
- 16 days
September 2026 Notes:02
Saw a thread today about how we’re likely in “the good ol’ days” of AI, referring to how relatively cheap these subscriptions are for the computing power we get, how it's unsustainable, and that it's only a matter of time before the cost-benefit ratio shifts the other way. Aside from price increases, we're already seeing things like models getting silently nerfed after they release so probably not too crazy of a prediction.
- 5 days
I spent 2 weeks optimizing my custom Claude Skills to be more token efficient when using Fable, only for Anthropic to release Opus 5.5 (which, from early testing, seems to be just as good for what I'm currently working on...)
- 33 days
August 2026 Notes:01
Most interesting find this week. Flipped a couple switches in my brain about my own AI workflow. The note about me (the human) being the bottleneck is something that's also crossed my mind recently. https://youtu.be/iQyg-KypKAA
- 98 days
May 2026 Notes:01
I notice I hesitate now to include em dashes in my blog posts to avoid having people assume my writing's AI-generated, despite using them regularly before they became AI markers.
- End of quick notes. 05 of 05.