Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

No offense but do you have any better use cases?

Automating posting AI written linkedin posts is basically littering the internet with trash as a service.



Anything more is hard. So it will need real engineers and visionaries.


No offense taken—LinkedIn-post bots aren’t the point. Think UNIX pipes for business ops: small, single-skill agents you chain, swap, and reuse.

Here are some concrete examples:

- Financial KPI radar pull-stripe -> calc-kpis -> write-digest — 2-hour spreadsheet grind becomes a cron job.

- Support triage classify-ticket -> fetch-answer -> route-edge — weekend inbox time drops 4 h → 30 m.

- Bug-fix conveyor parse-logs -> rank-severity -> draft-patch — Great for tiny teams with no dedicated QA.

- Receipts → Books ocr-receipt -> map-account -> push-p&l — zero-touch bookkeeping until tax season.

I'm still tuning the granularity—“one verb per agent” feels right so far. Thoughts?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: