Gamut
Recruiting & HR

Candidate Sourcing Agent

Source passive talent, validate every candidate on LinkedIn, and file them into your ATS with drafts ready to send.

Updated

Connects with

GreenhouseGreenhouseApolloApolloJuiceboxJuiceboxLinkedInLinkedInGmailGmailSlackSlack
Download template

The problem it solves

Sourcing for an open role eats hours: reading the JD, guessing the ideal profile, scraping a discovery tool, then manually checking each name on LinkedIn before drafting outreach. This agent runs that whole workflow end to end, profiling your current team to define the real candidate pattern and handing you a ranked, LinkedIn-verified finalist list with outreach already drafted in your ATS.

Who it's for

  • In-house recruiters and sourcers at growing companies
  • Talent acquisition teams at a recruiting agency or staffing firm
  • Heads of People or Talent owning multiple open reqs
  • Founders and hiring managers sourcing key roles themselves
  • RPO and contract sourcers running searches for clients

What it does

  1. 1

    Read the job description

    It ingests the JD you upload or point it to and extracts title, territory, experience, must-have background, and dealbreakers.

  2. 2

    Open the requisition internally

    It creates or reuses the req in your ATS and keeps it internal-only, verifying zero public job postings.

  3. 3

    Profile your current team

    It studies people already in that role to learn the real hiring pattern, background, tenure, and prior companies, not just what the JD says.

  4. 4

    Confirm the search profile

    It shows you a draft search profile of titles, locations, and company pool and takes your edits before sourcing.

  5. 5

    Source a calibration set

    It pulls a small first batch of candidates for your thumbs-up or thumbs-down feedback, which it uses to sharpen the full search.

  6. 6

    Validate finalists on LinkedIn

    It produces a ranked finalist list and opens each candidate's LinkedIn profile as the source of truth before advancing them.

  7. 7

    Draft two-channel outreach

    It writes a personalized LinkedIn message and email for every finalist in your recruiter voice.

  8. 8

    File everyone into your ATS

    It adds each finalist at the lead stage with both drafts attached as notes, marks them Reached Out, and stages one LinkedIn message in the browser, all draft-only.

Key benefits

  • Cuts a full sourcing pass from hours to a single supervised run
  • Every finalist is validated on LinkedIn before it reaches you, not just scraped from a database
  • Search profiles are built from your actual team's hiring pattern, so the list fits your bar
  • Your thumbs-up and thumbs-down feedback compounds and is remembered for the next search
  • Outreach is drafted on both LinkedIn and email and filed straight into the candidate record
  • Nothing sends without you, so you keep full control of every message

Sample use cases

You open a new Enterprise AE req in a competitive territory

The agent profiles your current AEs, sources passive sellers from similar companies, validates each on LinkedIn, and hands you three ranked finalists with outreach drafted in the ATS.

A recruiting agency needs to fill a client search fast without lowering the bar

The agent runs a calibration round, tunes the search to the client's feedback, and produces a finalist slate with personalized LinkedIn and email drafts ready for recruiter review.

Your discovery tool returns promising names but you've been burned by stale data

The agent treats the discovery engine as a first pass only and verifies tenure, prior companies, and domain on each LinkedIn profile, dropping anyone who doesn't hold up.

You want every sourced candidate documented for an audit trail

The agent files each finalist at the lead stage with a match rationale and both outreach drafts saved as notes, then marks them Reached Out under your source label.

Key integrations

  • Ashby, Greenhouse, or Lever

    Your applicant tracking system, where the agent opens the req and files candidates with notes and stages.

  • Apollo

    A discovery engine for finding passive candidates by title, location, and company pool.

  • Juicebox or MetaView

    Optional internal sourcing tools the agent can layer in alongside Apollo as additional candidate sources.

  • LinkedIn

    The source of truth for validating each candidate's tenure, background, and the channel for staging connection messages.

  • Gmail or your email account

    Connected via OAuth so the agent can create outreach email drafts, never sending them.

  • Slack

    Optional delivery channel where the agent can post a summary of the finalist slate.

The agent works the way a strong human sourcer does. It uses APIs where it can for speed and the browser where it must, since LinkedIn has no API, and it deliberately runs browser actions a little slower than a person to stay under bot protection. The first live run feels measured for that reason; once you trust it, it can run in the background while you work on other reqs.

Access is scoped tightly. The agent only writes to the one role's requisition and only sets sourcing stages, so it cannot touch other reqs or advance candidates toward interview or offer. Combined with the draft-only outreach, that keeps a recruiter firmly in control of every decision that matters.

Getting started

  1. Import the workspace — Add the Candidate Sourcing Agent to Gamut from the marketplace or by importing the workspace.
  2. Run the onboarding interview — A setup session walks you through your ATS, discovery engine, outreach channels, default role and territory, and candidate counts, then connects your accounts.
  3. Give it a role to source — Hand it a job description and say source it; it builds its sourcing skill on the first run and then works the full search live.

Frequently asked questions

Does the candidate sourcing agent send messages without my approval?

No. Every LinkedIn message and email is created as a draft only and filed into the candidate record. The Reached Out stage and the staged composer are your review-and-send point, so nothing ever goes out until you click send.

Which systems does the agent work with?

It connects to your ATS such as Ashby, Greenhouse, or Lever, a discovery engine like Apollo or an internal tool such as Juicebox or MetaView, LinkedIn for validation, and your email for drafts. You choose the exact tools during onboarding and the workflow stays the same.

How is this different from sourcing manually or with a generic tool?

A generic discovery tool just returns a list of names. This agent profiles your current team to define the real ideal-candidate pattern, validates every finalist on LinkedIn rather than trusting database data, and files outreach into your ATS, doing the whole workflow a human sourcer does end to end.

Will the role be posted publicly?

No. The requisition is kept internal-only by default and the agent verifies zero public job postings for it. This is built for passive sourcing, not for posting and waiting on inbound applicants.

How does it make sure candidates are actually a good fit?

It sources a small calibration set first and asks for your thumbs-up or thumbs-down on each, then tunes the full search to that feedback. The more specific your reasons, the sharper the final list, and it remembers your preferences for future searches.

Is the agent free to use?

The template itself is free to import into Gamut. You bring your own accounts, so any costs come from your existing ATS, your discovery engine subscription, and any API usage those tools charge for.