Google Jobs — Every Board's Openings, With the Pay Attached

August 18, 2026

Job hunting means opening eight tabs: LinkedIn, Indeed, the company's own careers page, three aggregators nobody has heard of. Google Jobs already pools all of them, and now Alfrada OS reads that pool directly. Ask "find me remote product manager roles" or "who's hiring data scientists in Berlin, and what do they pay?" and you get live listings with salary, schedule, posting age, the qualifications Google extracted, and a link that goes straight to the application — plus a CSV of the whole set with every full description in it.

What you can do

• Search the aggregate, not one board. google_jobs returns listings Google collected from LinkedIn, Indeed, company career pages, and job boards, each credited to the source it came from (via) with the other boards carrying the same role listed underneath.
• See the decision facts up front — salary range where the employer published one, full-time / part-time / contract, how long ago it was posted, remote flag, "no degree mentioned", and the benefits chips Google attached (health insurance, paid time off, dental).
• Read the requirements without opening the posting. Google's extracted Qualifications bullets come back with every listing; ask for full descriptions and you also get Responsibilities and the posting text itself.
• Filters go in the sentence, not in a form. Google Jobs parses them from the query: "remote", "part time", "internship", "no degree", "since yesterday", "in the last 3 days". Write the search the way you'd say it.
• Up to 30 listings per call. Ten by default; ask for a wider sweep and Alfrada OS pages through Google's cursor for you and de-duplicates across pages.
• Every search saves as a CSV. jobs_<query>_<timestamp>.csv lands in the session out/ folder with position, title, company, location, source, posted date, schedule, salary, remote flag, benefits, apply link, other boards, and the complete description text — indexed so Code Executor can filter it, Excel Ops can join it, and Report Generator can cite it. Pass save_data: false to skip.
• Local means local. Name a city and Alfrada OS makes Google actually search there — if Google doesn't recognise the place, it retries with the location written into the query rather than silently handing back results from the other end of the country, and tells you when a search ended up country-wide.
• A shortlist in the chat, not a JSON dump. The result card shows each role with its company logo, pay, schedule and posting age as chips, the qualification bullets underneath, and the whole card links straight to the apply URL.

Where this shows up

• Actually looking for a job. "Find me remote python developer roles posted in the last 3 days." Recency filter in the sentence, apply links one click away.
• Checking what a role pays before a salary conversation. "What are data scientist jobs in New York paying right now?" Salary ranges across ten live postings beat one glassdoor average from 2023.
• Sizing a competitor's plans. "Who's hiring machine learning engineers in London, and for what?" A company's open roles are the most honest roadmap it publishes.
• Building a shortlist you can work through. "Get me 30 product manager openings in Berlin and put the ones offering over €80k in a sheet." The CSV carries every full description, so the filtering happens in one turn without 30 more searches.
• Hiring-market research for a report. "How many companies are advertising for prompt engineers, and what do the postings ask for?" The extracted qualification bullets aggregate cleanly.

Try it

• "find me remote product designer jobs"
• "who is hiring data scientists in New York and what do they pay?"
• "part time barista jobs in Chicago posted this week"
• "machine learning engineer roles in London — show me the full descriptions"
• "30 python developer openings, then filter for the fully remote ones"
• "what qualifications are warehouse jobs near me asking for?"

Applying with it, not just searching

Finding the role is the easy half. Alfrada OS ships a job search playbook the agent opens on its own when you are the one job hunting, and it runs the rest of the chain: it looks for a CV you uploaded in this session or any past one, reads your public LinkedIn profile if you paste the URL, and asks — once — only for what is still missing (target title, location, contract type, salary floor, work authorisation).

Then it searches both boards — Google Jobs for coverage, LinkedIn Jobs for LinkedIn's own listings and filters — and scores each role for ATS fit: hard requirements you miss, the exact keywords the posting uses that your CV never states in those words, title and seniority alignment, and whether the file itself will survive a parser (multi-column layouts, tables, and image-exported PDFs are the usual reasons a strong CV arrives as gibberish). You get a ranked shortlist, the gaps costing you the most points, an edited CV — patched in place when it's a .docx, so your formatting survives — and paste-ready LinkedIn headline, About and skills text.

One rule it holds: missing keywords go in only where you genuinely have the experience. A gap is reported as a gap, never written into your skills section.

Reading someone else's hiring

Job postings are not only for job seekers. A company's open roles are it stating, in public, where the next money goes — so live listings work as an alt-data source for diligence, market sizing and competitor tracking, and for a private target they are frequently the only continuous operating disclosure available. Ask what a competitor is building and Alfrada OS breaks their postings down by function, seniority and location, mines the requirements for the technologies and compliance regimes they name, and reports the published pay bands.

The method matters more than the pull, and the alt-data playbook enforces it: both job tools return a top-N sample (30 listings per Google Jobs call, 100 rows per LinkedIn Jobs search), so a raw count is never headcount growth. What compares run to run is composition, net-new postings matched on their apply links, and pay — off a query set held fixed, the same discipline rank tracking needs. It also handles the five ways job data misleads: syndication double-counting, evergreen reposts that look permanently fresh, agency listings under the wrong company name, calendar-driven seasonal hiring, and a company that stops posting — which is a hiring freeze, not missing data.

The first run already carries history: postings state their age, so bucketing what is live today by posting date reconstructs roughly the last month of hiring flow before any collection has happened.

Heads up

• Google occasionally answers a good search with nothing. Alfrada OS retries once automatically; if a search you know should have results comes back empty, asking again usually returns them.
• Google Jobs coverage is uneven by country. The US, UK and Pakistan are dense; some EU cities return little or nothing for a given phrasing even though local job boards are full. An empty result is Google's index, not a failure — broaden the title or widen the location.
• A location Google doesn't recognise widens the search. Alfrada OS retries with the place in the query text and, if that also comes up short, says plainly that the listings are country-wide rather than passing them off as local.
• Descriptions are trimmed in chat, complete in the CSV. Postings run 2,000–6,000 characters each; the card and the model see a snippet plus bullets, and the saved CSV keeps the full text for anything that needs it.
• Salary appears only when the employer published one. Google does not estimate it, and it prints ranges the way the poster wrote them.
• Same rate as the rest of the SearchAPI family. One Google Jobs call costs the same as one Google News or Google Search call; a 30-result sweep is three calls.
• Different from LinkedIn Jobs. linkedin_jobs reads LinkedIn only, with applicant counts and LinkedIn's own experience-level filters. google_jobs is the aggregate across every board — reach for it first, and drop to LinkedIn when the question is specifically about LinkedIn.