Every recruiter carries a version of this around in their head. The strong candidate from a process two years ago who lost out to someone marginally better. The person a colleague met, liked, and wrote three sharp notes about. The applicant who wasn't right for that role but would be perfect for this one.
Software has never been able to use any of that. It matches keywords against CVs and calls it search. The knowledge your team actually accumulated sits in the system doing nothing.
We rebuilt candidate search and screening in Vouch around a different idea: everything your team has ever learned about a candidate should count, every time you search.
Two teams running the identical query get different results, because their histories are different. That's the point.
How it works now
You describe the role the way you'd brief a colleague. The skills and experience you need. Where the person should be. And just as important, what you want to avoid: the seniority level that never works out, the profile type that looks right on paper but isn't.
Vouch evaluates your entire pool against that brief, weighing every layer of context it has on each candidate:
- What's on their profile: experience, education, languages, skills
- What your team wrote: notes, comments, and evaluations from earlier processes
- What actually happened: which processes they were in, how far they got, past placements
- What they've said: interview transcripts and meeting history, when available
You get a ranked list, and each candidate comes with a plain-language explanation of why they placed where they did. Their strengths against your brief, their gaps, the reasoning. You can agree or disagree, and when you rate the results, the system learns what you mean.
This is the difference between software that searches a database and software that remembers. Your ten years of placements is now an advantage no new tool can copy.
The same brain, screening your inbound
The rebuilt evaluation doesn't only run when you search. When candidates apply through Finn, Arbeidsplassen, or The Hub, each application is evaluated against your specific brief for that role, automatically, before you've opened a single CV.
That means the 80 applications waiting on Monday morning arrive pre-read: ranked against your actual criteria, with reasoning attached, applied the same way to applicant number 1 and applicant number 80. You still make every decision. You just stop spending your best hours discovering which CVs weren't worth opening.
Language is no longer a filter
A quiet failure mode of every search tool: the great candidate whose profile is written in a different language than your query. They simply never show up.
Vouch now expands searches across six language families. A Norwegian query finds candidates whose profiles are in English, Swedish, Finnish, or Dutch, and the reverse. If you hire across the Nordics or recruit international talent, your search results just got bigger without your pool changing at all.
Built to evaluate candidates from anywhere
The evaluation model doesn't care where a candidate came from. Today it runs on your talent pool and your inbound applicants. As new sources connect to Vouch, the same model will evaluate those candidates too, against the same brief, with the same explanations. The screening layer is built once, and everything flows through it.
The screening layer is built once, and everything flows through it.
Also in this release
The rest of the release is substantial in its own right, so briefly:
- A new Analytics section showing candidate flow through your stages, time to offer, time to hire, average time per stage, and which channels bring applicants that convert.
- Verified academic credentials: candidates share verified grades through Vitnemålsportalen directly in the application form, via Feide or ID-porten. Norwegian institutions are live, with the groundwork in place for the rest of the Nordics.
- Candidate AI opt-out: applicants can decline AI review, as GDPR gives them the right to. Their application routes to manual review, with a clear flag for recruiters.
- Enforced two-factor authentication, workspace-shared templates and booking links, scheduled bulk sends, candidate references in the application form, and interview stages you can import from existing positions.
The full list is on our changelog. But if you only try one thing: open search, describe a role you know well, and see who your own history surfaces.