Candidate Checks

What a candidate check means

A candidate check means checking one individual candidate against one specific briefing.

Credits

Credit usage depends on your LLM setup and whether the candidate was checked recently.

• Billing factors:
  • LLM mode: TalentSourcer AI managed or BYOK
  • Check type: new check or existing recent check

Important: recent checks are network-aware. If someone in the TalentSourcer AI network already checked that candidate within your recency window, your check can use existing-check pricing.

Cheeky but true: more users in TalentSourcer AI means a higher chance candidate checks are already recent, which can save your team credits.

Candidate recency window (credit control)

Recency window controls when a candidate is still treated as recent for pricing.

• Options: 30, 60, or 90 days
• Higher window means more candidates count as recent
• More recent candidates means lower average credit usage

You can adjust it here. (Open Candidate Recency Window)

What BYOK means

BYOK (Bring Your Own Keys) means you connect and use your own LLM provider key (OpenRouter, OpenAI, Anthropic, or Google) instead of TalentSourcer AI managed AI.

Why teams choose BYOK

• You get more candidate checks out of the same subscription spend.

• BYOK usually lowers overall check cost because managed pricing needs conservative buffers for variable data complexity.

• You get more steering power over quality vs cost by choosing model and provider per use case.

• You can run premium models for high-stakes roles and cheaper models for high-volume roles.

• Configure provider keys here. (Open LLM Provider Settings)

• For full setup and cost tracking guidance, use this page. (Open BYOK and AI Cost Controls)

Triggering candidate checks

1. Open your project search and select candidates.
2. Click Check Candidates.
3. Select the briefing version in the dialog.
4. Review credits needed.
5. Click Start Check.

For larger runs (100+ candidates), enable Notify me when all selected candidate checks are done.

Use Select all candidates in this search carefully. It can trigger large credit usage in one run.

What happens behind the scenes

For each candidate check, TalentSourcer AI:

• Uses LinkedIn profile data for the candidate
• Adds research on current and past employers
• Evaluates every must-have and nice-to-have requirement as yes or no, with reasoning
• Calculates fit score and assigns category

Reviewing candidate checks

Recommended review flow:

1. Use View by Briefing.
2. Filter by category (A, B, C).
3. Sort by fit score.
4. Open individual checks and review requirement reasoning plus employer research.
5. Open the candidate's LinkedIn profile for spot validation.

Candidate categories (same definitions as in app)

• A Candidates: Fulfill all must-have requirements and all nice-to-have requirements
• B Candidates: Fulfill all must-have requirements and some nice-to-have requirements
• C Candidates: Fulfill all must-have requirements but no nice-to-have requirements
• No Fit Candidates: Lacking at least one must-have requirement

Start small when briefing is new

If the briefing is still fresh, do not start with thousands of checks.

• Start with 10 to 15 candidates
• Review outputs in detail
• Cross-check the real LinkedIn profiles
• Confirm requirements are clear and complete before scaling

If pool size is off after checks (too many or too little candidates)

If you do not have enough strong matches, or you have too many weak matches, use Talent Pool Insights and recategorize.

Find the walkthrough here. (Open Talent Pool Insights Guide)