How to describe the job you want so an AI can actually find it

Why does "senior developer, remote" find so little?

It looks specific. It has a seniority word and a location word. But it can't tell a backend Go role from a frontend React role. It can't tell a "remote" that means anywhere from a "remote" that means one country. It says nothing about salary or dealbreakers. Every word in it is true of thousands of postings, so none of the words filter anything.

A useful brief is a spec, and a spec can be wrong. Each clause should help rank or rule out a posting when the source gives enough evidence. "Senior" can conflict with an explicitly junior role. "Remote, EU-based" can conflict with a US-only listing. "$100k+" can conflict with a stated $70k ceiling. Model-provider guidance says the same thing about prompts: be clear, direct, and specific about constraints. It doesn't mean a model can recover facts the posting never gave it.

This is the same idea as scoring a posting before you apply instead of after. I wrote about it in why 200 applications can return zero responses. The brief is the input that scoring runs on, and a vague brief gives any scorer fewer honest ways to tell two postings apart.

What goes in a job brief?

I think five things carry almost all of the signal. Get these right and the rest is polish.

1. Stack or domainNot "developer". "Go backend", or "growth marketing at a B2B SaaS". The stack tells a scorer what the job actually is. A title alone tells it very little.
2. Seniority"Senior" means something different at a ten-person startup than at a bank. Give a level or how much you have owned, not just the word.
3. Location eligibility"Remote" is a claim, not a location. Say "EU-based", "US only" or "CET±3" so a quietly restricted listing can be caught. Here is [how that restriction hides](/articles/remote-jobs-eu-location-eligibility/).
4. Salary floorA number, not a hope. A posting below it wastes both sides' time. A posting with no range is an unknown, and a good scorer should say so instead of guessing.
5. DealbreakersNo crypto, no agencies, no on-call, no five days in the office. When a posting states the opposite, these can rule it out outright.

Why are dealbreakers the most useful words?

Here is the part that is easy to underweight. Positive terms rank. Dealbreakers filter, and filtering is where most of the saved time is.

"React" or "growth marketing" tells a scorer what to prefer among postings that are still in play. It is soft, because a posting missing one preferred skill can still be a good match. "No crypto" does something different. It doesn't take a point off a score, it removes a whole category in two words.

What does a sharp brief look like?

That feeling is where Oink came from. Three vague sentences next to what I would write instead:

The information is the same every time: stack or domain, seniority, where you can work, a number, and what rules a posting out. The exact words matter less than the shape. Swap in your own stack, city and floor and it still works.

Where do you use it?

You don't need a tool. Write the five-part sentence once, keep it in a note, and paste it into whatever board you are searching that day. It even works as a rubric you score by hand in a spreadsheet, one line per posting, five minutes a day.

It stops being manual when the volume does. In Telegram, a preference brief can start the profile and a CV is optional. Oink then searches the sources available for that run and ranks the viable ones from 0 to 5, with the reasons it fits, its concerns and what is unknown, all tied to your brief.