What we observed
On October 18, 2026, we requested five postings each from Ramp, Notion and Linear. The Actor successfully read all three boards, which reported 324 postings in total, and saved the selected 15. These are the first retained records within each board’s response order, not a random or newest-first sample.
5 of the 15 saved postings had both numeric salary bounds. All five were in the Ramp slice. The five Notion and five Linear observations had no structured numeric range in this output. That does not establish that the employers omit salary elsewhere on their job pages. We did not collect or search full job descriptions.
A second signal worth keeping
2 saved records had workplaceType set to Hybrid while remote was true. The Actor can identify a remote location option on a posting whose overall workplace label is Hybrid. Treat these fields as separate evidence: open the employer page and check the applicable location before calling a role fully remote.
What this changes in a review sheet
Keep salaryMin, salaryMax, salaryCurrency and salaryInterval together. Preserve a missing value as missing; do not turn it into zero or compare it with a disclosed range. Use company plus jobId for record identity, and retain jobUrl so a reviewer can inspect the original posting.
The small sample demonstrates a useful extraction distinction. It does not estimate market-wide salary transparency, compare employer hiring quality, establish compensation for every location, or measure changes over time.
Reproduce the observation
Use the downloadable input and the linked workflow. The same input may return different postings later because source order and vacancies change. The saved sample is fixed to the observation timestamp, and the source links may later expire. Five postings per board and 15 total were intentional caps.
Original board sources: Ramp, Notion, and Linear. Individual posting links are preserved in the downloadable evidence.