> ## Documentation Index
> Fetch the complete documentation index at: https://docs.needle.so/llms.txt
> Use this file to discover all available pages before exploring further.

# Sourcing

> Have the agent find and score candidates for a job.

Sourcing runs as a conversation: you describe who you're looking for, the agent searches and scores, you review the results.

## Start a search

Open the job's **Sourcing** view and describe the profile in the chat, the way you'd brief a recruiter — the experience that matters, location, anything to avoid. The agent turns your brief into explicit criteria that you can check and correct before the search runs.

## What the agent does

It searches professional data sources and the open web, then evaluates every profile it finds against the criteria. Each candidate gets a score and a short explanation tied to their background, so you can see *why* they ranked where they did.

## Review the results

Found candidates appear in the Sourcing view, ranked by score. Go through them and accept the ones worth pursuing; pass on the rest. Your decisions feed back into what the agent looks for next.

## Refine mid-search

If the results show you the brief was off — too many big-company profiles, wrong seniority — say so in the chat. The agent updates the criteria and re-scores; you don't start over.

## After the shortlist

Accepted candidates are ready to contact. The usual next step is enrolling them in an [outreach sequence](/outreach).