Step 2 · Market search
Source passive candidates with AI, without building every search by hand
AIcruiters is a candidate sourcing tool that runs the search for you. Starting from the reviewed vacancy strategy, the agent explores several search routes across available talent sources at the same time, removes duplicate and irrelevant profiles as it goes, and delivers a focused long list that is ready to be ranked against the full vacancy.
Why is manual candidate sourcing so slow?
Manual sourcing is a loop. A recruiter writes a Boolean string, scans the first pages of results, opens profiles, notices the search is too broad or too narrow, rewrites the string and starts again. Each variation costs time, so many searches stop after one or two routes. Candidates who describe the same experience in different words never show up.
The second problem is noise. Long result lists contain people already contacted last month, profiles that match a keyword but not the job, and the same person found through two routes. Clearing that noise is invisible work that eats a large part of a sourcing day. Our comparison of AI and manual sourcing sets out where each approach works best.
How does the agent search the market?
The agent does not rely on one query. It builds several routes from the signals in the vacancy analysis and runs them in parallel, so candidates who describe their work differently are still found. Which routes are used depends on the vacancy; typical examples:
- Title and skills route: the official title, its variants and the core skills
- Career path route: people in the roles that typically lead to this one
- Industry route: the same work in adjacent sectors, under a different name
- Clean-up as it goes: duplicates merged, clearly irrelevant profiles removed
- 1,240Profiles found across talent sources
- 312Relevant after de-duplication and filters
- 25Ranked against the full vacancy
- route 01 · title + skills684
- route 02 · career path391
- route 03 · industry + stack165
- duplicates removed−188
- irrelevant removed−740
What makes a search route better than a Boolean string?
A Boolean string is a filter: a profile either contains the words or it does not. That works when the job has a stable, well-known title. It breaks down when the right people use other words, when the title is new, or when the best candidates come from a neighbouring field. Writing strings that cover every variation quickly becomes unreadable, which is why experienced sourcers often run several short strings instead of one long one. Our Boolean search guide covers the syntax.
A search route follows the same logic, but the agent generates and runs the variations itself. Each route has its own purpose, and the results are compared and merged. The recruiter does not have to decide upfront which single string is best.
Where manual search still makes sense
A recruiter who knows exactly who they want, for example a named person or a very small group at a few companies, is often faster by hand. Referrals and your own network remain sources the agent does not replace. Agentic search pays off when the job is recurring, the market is broad, or the right people are spread over many titles and sectors.
How does AI sourcing differ from a search seat?
A search seat, such as LinkedIn Recruiter, gives recruiters access and filters. An agent takes over the execution. Both have a place; the question is who does the repetitive work.
For a full comparison of costs and workflow, see AIcruiters vs LinkedIn Recruiter.
What does the recruiter see after the search?
Not a raw result list. The search produces a long list of relevant profiles, and that list goes straight into candidate ranking, where each person is scored against the complete vacancy with visible evidence. When the list is full of the wrong audience, the recruiter adjusts the strategy, not a string.
If the long list is thin, that is information too. It usually means a hard requirement is too strict or the market for this profile is small. The recruiter decides whether to widen the strategy, and the agent searches again.
Where does AI candidate sourcing stop?
The agent searches available talent sources within the setup agreed with your team; it does not decide on its own which new sources to use. Which sources are part of your workflow, and how candidate data is accessed, used and retained, is set per workflow and discussed during the technical and privacy review.
Approaching people who never applied also has privacy implications. Under the GDPR, a person whose data you did not collect from them directly must be informed, at the latest when you first contact them. Our guide to GDPR and candidate sourcing explains what that means in practice, and the security page describes how data control works in AIcruiters.
Questions about AI candidate sourcing
Does it find passive candidates?
Yes. The agent searches for people who fit the vacancy, not for people who applied. Most of them are not actively looking, which is why the outreach step is personal rather than generic.
Which talent sources does AIcruiters search?
The agent works with the talent sources available in your agreed setup. We go through the sources, access and data handling for your workflow during the technical review, rather than promising a fixed list here.
Does it replace our Boolean searches?
For recurring and broad searches, largely yes, because the agent builds and runs the variations. For a very targeted search on a handful of named people, a recruiter by hand can still be quicker.
Can we exclude people we already approached?
Duplicates are removed during the search. How previous contacts from your own systems are taken into account depends on your setup and ATS, which we assess per ATS.
How big is the database?
AIcruiters is not sold as a database with a profile count. It is an agent that searches, filters and ranks, so the useful number is the relevant long list for your vacancy, which you see in the demo.
Watch the agent search for one of your open roles
Bring a vacancy you are sourcing for right now. You see the search routes, what gets filtered out and the long list that goes into ranking.