Applicants & Pipeline

How applicants are screened and ranked, and how your team moves them from applied to hired.

What happens when someone applies

1. The application is captured (resume upload or LinkedIn URL, plus any apply-question answers).
2. Knockout rules are evaluated first. A disqualifying answer declines the applicant instantly — no AI call — and (if enabled) sends a polite rejection.
3. Otherwise the applicant gets a confirmation email with their private status link.
4. The resume is parsed into structured fields.
5. An AI fit score is produced against the job's calibration.
6. The applicant appears in the job's pipeline, ranked by fit.

Screening is automatic and bounded by your plan's monthly applicants screened limit and per-job/day caps.

Reading an AI fit score

Each scored candidate shows:

• Score (0–100) and confidence (drops when candidate data is thin),
• Summary, strengths, gaps, missing data, and risks,
• Evidence — short claims each tied to a specific candidate field (so you can trust and verify the score, not just take a number).

Scores are reproducible: the provider, model, prompt version, input hash, output, and cost are recorded for every run.

The pipeline (stages)

Candidates move through: applied → shortlisted → interview → offer → hired, or rejected. The board is a kanban with one column per stage; a card shows the candidate, an origin badge (Applied via careers vs. Sourced), and the fit score.

Move candidates with the card's forward/back arrows, the stage selector in the review drawer, or the sticky Advance / Reject footer. Moving to reject, interview, or offer opens the candidate-notification step (see below).

Screening rules (opt-in automation)

Off by default — hiring stays human-in-the-loop. Under Settings → Screening Rules an admin can enable:

• Auto-shortlist applicants scoring at or above a threshold.
• Auto-reject applicants scoring below a threshold (sends a polite decline).

These apply only at intake (the applied stage).

Reviewing efficiently

• Ranked shortlist — applicants are ordered by fit, strongest first.
• Comments — discuss a candidate with your team.
• Scorecards — structured competency ratings and a recommendation; the review drawer nudges you to collect enough scorecards before advancing.
• Hiring decisions — advance / reject / offer / hire with a required rationale (recorded).
• AI interview guide — generate per-candidate focus areas and questions.
• Scorecard consensus — synthesizes scorecards, comments, and AI context into a recommendation.
• Async pre-screen — if the job has pre-screen questions, candidates answer them from the status page and the answers fold into the fit evidence.

Bulk actions

• Bulk stage move — move up to 50 candidates at once.
• Bulk rescore — re-run AI scoring (e.g. after changing calibration).
• Bulk reject — decline up to 50 candidates, optionally sending a templated email.

Talent pool

Save strong candidates you're not hiring right now to the talent pool. When you open a new role, Kashif suggests talent-pool matches based on skill overlap and prior scores, so good people don't get lost between roles.

Candidate withdrawals

A candidate can withdraw their own application from the status page. When they do:

• their application is marked withdrawn and the pipeline entry moves to rejected,
• the card shows a red "Withdrawn by candidate" badge and the review drawer shows a banner explaining it was the candidate's choice (no action needed),
• the entry is locked — recruiters can no longer move it through the pipeline (only remove it), because the candidate owns that decision.

Sourcing (supporting flow)

Beyond inbound applications, you can search the web for candidates (Serper/Brave), enrich profiles from a LinkedIn URL (Apify), and add them to the same pipeline — where they're scored and reviewed identically. Sourcing is optional and not the primary motion.