HackerTrans
トップ新着トレンドコメント過去質問紹介求人

mondo_daemon

no profile record

投稿

Life lessons and hot takes from my 30s

shreyasprakash.com
2 ポイント·投稿者 mondo_daemon·13 日前·2 コメント

My agentic engineering workflow as someone who doesn't write code

shreyasprakash.com
4 ポイント·投稿者 mondo_daemon·22 日前·1 コメント

L2 Fat Marker Sketches

shreyasprakash.com
1 ポイント·投稿者 mondo_daemon·10 か月前·1 コメント

コメント

mondo_daemon
·22 日前·議論
When it comes to agentic engineering, every day is a year. There is always a new tool, library, new TUI apps, new language models, new benchmarks which topple the previous ones, etc etc. You name it.

Not just that, every other day, a language model which you've slowly started developing model intuition is now "dead"... making you scramble for the next shiny new toy to play with.

To bring some sanity to this process (and to my own thinking), I tried to document the "snapshot" (as of June 2026) of what my current agentic engineering process looks like:
mondo_daemon
·10 か月前·議論
What is the right fidelity to communicate one's ideas?

Too crude, and you might more honest but vague feedback.

Too detailed, and you might less honest but more tangible feedback.

Is there a spectrum here? I talk about how I've been exploring switching between fidelities for best representing ideas here in this essay