HackerTrans
TopNewTrendsCommentsPastAskShowJobs

Absonsonson

no profile record

Submissions

Whistle

abson.bearblog.dev
5 points·by Absonsonson·2 個月前·3 comments

Show HN: Unlegacy – document everything, from COBOL to AI generated code

unlegacy.ai
9 points·by Absonsonson·3 個月前·1 comments

Documentation is all you need - or how I stopped suffering with legacy code

blog.unlegacy.ai
4 points·by Absonsonson·3 個月前·1 comments

comments

Absonsonson
·2 個月前·discuss
Spread that cheer!
Absonsonson
·2 個月前·discuss
Stretch and try to touch my toes
Absonsonson
·2 個月前·discuss
Play guitar
Absonsonson
·3 個月前·discuss
Author here. In corporate engineering, nobody works from the same ground truth. Everything is planned on top of "ifs" and "I thinks." The person who actually knows is always in a meeting, on vacation, or just left and the next person has to redo the entire detective work from scratch.

The understanding never accumulates, it just resets. Does anyone actually solve this or do we all just accept the endless rework as normal?
Absonsonson
·8 個月前·discuss
Interesting approach — treating ads as something you can reverse-engineer instead of just “inspire from” is the real value here. Most creative tools generate content blindly; this actually tries to deconstruct what makes a winning ad work: structure, pacing, hooks, messaging patterns, etc.

If you can consistently break an ad down into reusable components (hook, setup, proof, CTA, visual pattern) and let people remix those templates, that’s powerful. It basically turns high-performing ads into a library of “design patterns” for creatives.

Curious how deep the reverse engineering goes: • Are you analyzing just text, or also visual rhythm, shot sequence, and narrative beats? • Can the system infer why an ad might be converting well, or just what it contains? • How do you avoid overfitting on trends from one platform/vertical?