Case study · 002 · TalentSupply
From chasing job orders to creating demand.
TalentSupply ran reactive recruitment: a job order arrived, and three to five agencies raced to fill it. Nucleotto built a signal-to-float pipeline that finds companies hiring directly, briefs the recruiter every morning, and keeps a human decision on every send.
TalentSupplyWhere they started
TalentSupply ran traditional reactive recruitment. A job order arrived, and three to five agencies raced to fill it. Prospecting meant manually scanning Seek and LinkedIn to guess who might be hiring.
Outreach was written cold, from scratch, each time. Strong candidates sat un-marketed because floating them was a manual chore. Everything lived in one inbox and one head.
What was getting in the way
What we saw could change
The best moment to act is when a company posts its own role on Seek: proven intent, and no agency in the mix. That is the float window.
How the work ran
Inside the workflow: how orders arrive, where the hours go, and where the margin dies.
Walks through how job orders arrive and how prospecting is done.
Finds the highest-value moment: direct-hire signals with no agency engaged.
Signal map and 97-company watchlist.
Build around the float window.
A signal-to-float pipeline: detect direct-hire signals, rank hot or warm, surface contact and angle, float an anonymised candidate.
Confirms the daily workflow and the rules for ethical floating.
Designs the pipeline, the daily cockpit, and one shared candidate filter.
Pipeline blueprint, clone-able per recruiter.
Approve the build scope.
A nightly scanner across 97 companies, a morning brief, a float engine, and a mobile-first cockpit.
Provides the watchlist and the candidate pool.
Builds the scanner, the ranked brief, the float engine, and the authenticated dashboard with an AI layer.
Working pipeline on live Seek data.
Run it nightly.
Verified end to end before the recruiter relies on it.
Reviews hot and warm calls by hand.
Verifies the scanner against live data, locks the API contract, and proves the dropdown and engine share one filter.
Tested pipeline and go-live checklist.
Go live.
A brief every morning. The recruiter decides every float.
Works the morning brief and decides who to float, and when.
Monitors scan freshness, tunes ranking, ships improvements.
Live daily system with stale-scan flagging.
Extend to the next recruiter.
What changed
How we kept it controlled
The system surfaces and drafts. People decide. Six mechanisms hold that line.
Nothing sends autonomously. The recruiter decides who to float, and when.
Candidates carry no names in the pool.
The shared filter blocks anyone already placed from an accidental float.
Every route sits behind auth.
AI outputs are drafts. A person edits and publishes.
Stale scans are flagged before the recruiter acts.
What it returned
Where it can go next
TalentSupply enters the market before the agency race starts.
Nucleotto built the daily system behind that. The recruiter holds every send.