Skip to content

Normalize a page of job listings

中文 · All use cases

Fifty listings, each hiding the salary and remote policy in a different paragraph. Dassi reads them once and gives you the same five columns for every one.

New to Dassi? Follow the installation guide and model setup guide.

Open the job search results page in Chrome (LinkedIn, Indeed, a company careers page). Open Dassi alongside it. Start with the listings visible on one page.

Read the job listings on this page. For each, give: title, company, location, remote policy (remote / hybrid / onsite / not stated), salary range as written (or "not stated"), years of experience required, and the 3 main skills named.
Read from the listing text only; write "not stated" rather than inferring. Don't open the listings or apply to anything.
From that table, hide anything that requires more than [N] years or is onsite. Rank the rest by how well they match [my stack / my priorities] and explain each ranking in one line.

Open two listings from the shortlist and compare the salary and years fields to the listing — those are the fields sites format most inconsistently. “Not stated” should be common; if Dassi fills every cell, it is guessing. The table is a snapshot of one page; paginate by repeating step 1.

Try another use case