Sourcing · In-house TA, niche tech roles
Rerun a search every week without paying for the same people
Keep a role warm. Reruns skip pages you already paid for and people already in your seen register.
Who it is for
In-house talent teams with hard-to-fill technical roles, where the market is thin and new people appear a few at a time.
The problem
Most tools charge again for the same results, and the same names come back every time.
Workflow
- 01
Mark
Delivered candidates go into the seen register automatically, or through set_candidate_status for people you sourced elsewhere.
- 02
Rerun
A new run on the same plan resumes from checkpoints. A page already paid for is free.
- 03
Extend
extend_search raises the target with an estimate clamped to the budget you add.
- 04
Check
check_candidates answers "have we seen this person?" for a list of profile URLs or name and employer.
Sample transcript
- You: Rerun last week's Ghent backend search. Only new people.
- Tool call: extend_search { run_id: "run_3Fx1", add_target: 20, add_budget_credits: 6 }
- Result: ← estimate within 6 credits · seen people skipped · requires_confirm true
- You: OK.
- Result: ← 2 new candidates · duplicates_skipped reported on the run
Outcome
- Only new people each week, with duplicates counted on the run.
- No second charge for a page you already bought for that query.
Example prompts
Weekly rerun
“Rerun last week's Ghent backend search. Only new people, and stay within 6 credits.”
Already in our ATS?
“Mark these 30 profile URLs from our ATS as seen so they never come back in a search.”
Check before you look
“Have we seen any of these people before? Check them against the seen register.”
Tools and endpoints
| tool | endpoint |
|---|---|
| extend_search | POST /v1/runs/{run_id}/extend |
| set_candidate_status | POST /v1/seen |
| check_candidates | POST /v1/seen:check |
Credit cost
example data| Pages already paid for | 0 |
|---|---|
| 2 new leads × 1 | 2 |
| 2 new live profiles × 2 | 4 |
| Total | 6 |
Example data. A quiet week can cost nothing at all.