AI sourcing that turns a brief into a shortlist
Sourcing usually breaks into four steps: describe the person you need, find plausible profiles, check that the details are real, and get them into a pipeline without retyping anything.
This page explains how an AI sourcing layer handles those steps, where the browser extension fits, and which decisions stay with the recruiter.
What the sourcing layer actually does
The workflow has two entry points. Inside the platform, you describe a candidate in ordinary language or switch to structured filters for company, title, keywords and location. If typing is the bottleneck, voice input accepts a spoken search instead. Outside the platform, a Chrome extension captures candidate, contact and company records while you browse supported sites.
Either way the record lands in the same database. Profiles arrive enriched with summaries, work history, education and location, and you can push a chosen profile into the system with a single click rather than copying fields into a spreadsheet. Candidates can then be attached to several open roles at once, and duplicate matching keeps repeated records from piling up as volume grows.
Two search modes, and when each one earns its place
Plain-language search is fastest when the brief is still fuzzy — you know the shape of the person but not the exact title or keyword set. Structured Boolean-style filters are better once you have a precise target and want repeatable, auditable criteria across a team.
A practical split: use natural language for the first pass on a new role, then convert what worked into filters so colleagues can rerun it. Voice input is a convenience for the same search, not a separate capability — treat it as an input method rather than a quality lever.
How a shortlist gets built: a worked example
Suppose you are filling a mid-level operations role. You run a plain-English search, and the results page returns up to 100 enriched profiles, each carrying a summary, work history, education and location. You scan for fit, discard the obvious mismatches, and import the ones worth a conversation. From there you assign the same candidate to two or three related openings rather than re-sourcing for each.
Note the two different numbers in that flow. Enrichment returns up to 100 profiles per search page. A separate review view presents up to 20 matches per page for a quality decision. They describe different screens, so do not read one as a cap on the other.
The browser extension: capture while you browse
The extension is for the moments when you find someone in the wild rather than through a search. It captures profiles from LinkedIn, Sales Navigator, Outlook, ZoomInfo, Xing, Indeed, Naukri, SEEK and other supported platforms, pulling available work history, education, skills and resume information without manual entry. You can also create notes, tasks, meetings, jobs and hotlists while you source, so context is not lost between the find and the follow-up.
Decision checklist before you adopt
Use this to test whether a sourcing tool fits your team, not just whether it has the feature list.
- Does search accept both free-text briefs and structured filters, or only one?
- Are profiles enriched at the point of search, or do you enrich after import?
- Is import one click into your own database, or an export file you re-upload?
- Can one candidate be attached to several jobs without duplicate records?
- Does browser capture cover the platforms your team actually uses?
- Do you know how many profiles appear per page, and how many you review per page?
- Is the compliance position documented, including the data processing agreement?
If the answer to the first three is "export and re-upload", the tool is a search engine, not a sourcing workflow.
Wrong fit versus duplicate: two different failures
These get conflated, and they need different fixes.
A wrong-fit result is a relevance problem. The search returned profiles that do not match the brief. The remedy is a tighter brief, a switch to structured filters, or a different phrasing — the recruiter still decides who enters the pipeline, so a poor result costs review time, not data quality.
A duplicate is a data problem. The same person exists twice because they were captured from two sources or imported twice. Duplicate matching addresses this at the database level, which matters more as sourcing volume rises. Fixing relevance will not fix duplicates, and vice versa.
Where the recruiter stays in control
The AI finds; you decide. Every profile returned is enriched and ranked against the brief, but the choice of who moves forward remains a human review step. That division is the reason a sourcing tool can speed up the top of the funnel without taking over candidate judgement.
It also sets the expectation for evaluation: judge the tool on how quickly it gets a credible set of profiles in front of a recruiter, and on whether the records that follow are clean enough to work with.
Frequently asked questions
Does AI sourcing replace a recruiter's judgement?
No. The search returns enriched profiles ranked to your brief, and you review up to 20 matches per page before deciding who enters the pipeline. The finding is automated; the quality call is not.
How is this different from sourcing on LinkedIn and copying profiles across?
The difference is the handoff. Instead of copying profiles into a spreadsheet, sourcing, enrichment and import happen in one flow, candidates can be assigned to jobs, and records stay synced in the same database.
What compliance position should I check before sourcing?
Ask for the GDPR position, the security audit status, and the data processing agreement that covers how records are handled. Sourcing from publicly available professional data is the stated basis; confirm the current documentation with the vendor before rollout.
How many profiles come back per search?
Up to 100 enriched profiles per search page, each with summary, work history, education and location. The review view for quality decisions shows up to 20 matches per page — a different screen, not a conflicting limit.