Rediscover the candidates you already own before you buy more

Every large candidate platform restricts automated access and bulk extraction in its terms of service, and enforces it through rate limits, detection and account suspension. Some offer a sanctioned API for partners; others reserve automated access to their own paid products. A tool that scrapes a network from your logged-in session is not a clever workaround, it is a way to lose the seat licence your desk depends on, and the consequence lands on you rather than on the vendor who shipped the scraper.

So the rule we work to is simple. Where a platform offers a sanctioned integration path, we take it and we describe the integration by its actual state: Live, In development or Roadmap, with no implied partnership and no logo we do not have rights to. Where a platform does not, we do not build a covert one, and we say why on the integration page rather than leaving a coming soon tile up for two years.

An agency database of forty thousand records typically has several hundred people in it who match a live brief right now, and no realistic way to find them. The cheapest placement you will ever make is somebody you already spoke to, whose consent you already hold, whose notice period you already know.

That is what ships today. Matching scores your existing candidates against a live requisition with a requirement-level breakdown. Reverse matching ranks open requisitions for a given candidate, which is the query you need when a strong person comes off a finished search. Saved searches turn a brief into a live list that repopulates as records arrive. Talent pools hold the lists a desk works every week. None of it requires anybody's terms of service, because the data is yours.

The second sourcing path is capture at the point where a recruiter is already looking at a profile they have legitimate access to. The browser extension pulls the visible details into a candidate record, checks for a duplicate before it writes, and attaches the requisition it was captured against.

The distinction between that and autonomous crawling is not cosmetic. A recruiter acting inside a product they hold a licence for, capturing a record they are looking at, one at a time, is a different activity from a background process enumerating a platform at machine speed. The first is how sourcing has always worked. The second is what the terms are written to stop.

When the Candidate Sourcing Agent arrives in Wave 2, its first and largest job will be your own database: reading a new requisition, searching the whole record set, and surfacing rediscovered candidates with the reason each one surfaced, before any external sourcing spend is considered.

Beyond that it will work through sanctioned channels only. Where a job board or network offers a partner search API, the agent will use it under whatever quota and attribution rules that partner sets. Where a platform does not, the agent will not go there, and the interface will say so plainly rather than returning quietly thinner results. It will also assemble the shortlist as a proposal for a recruiter to approve, in the same pattern as every other agent here.

A sourced candidate is a person whose data you now hold, and the obligations start at the moment the record is created rather than at the moment you contact them. Every record created by sourcing will carry its source, the date, the requisition it was captured against and the consent basis in force.

Retention applies from that point. A sourced record that is never contacted and never responds should not sit in a database indefinitely, and the retention window flags it for review or erasure on the schedule you configure. Where a jurisdiction requires notice to a person whose data you have collected, the record holds what you need to issue it. Issuing it remains your responsibility as the controller of that data.

Matching and shortlisting against your own database, saved searches and talent pools, the browser extension for capture, multi-board posting for inbound, and the referral and rediscovery routes that most desks under-use. For most agencies the honest answer is that the database they already own is under-searched, and an autonomous external agent would be solving the second problem while the first goes untouched.

If external sourcing automation is decisive for your evaluation, weigh it as a Wave 2 item and ask about it in a demo. Progress appears on the shipping page and the roadmap, and items move between waves in public.