Recruiting workflows often become a chain of tabs, copy-paste and decisions nobody can explain later. The prompts below are designed to force evidence into the process.
They are not magic words. The useful part is the structure: each prompt names the input, the decision and the evidence required.
The full open playbook is available on GitHub.
1. Turn a vague brief into search criteria
Convert this hiring brief into a structured sourcing plan. Return target titles, adjacent titles, must-have skills, likely false positives, target locations, target companies and five questions that would materially improve the search. Brief: [paste the real brief]
This catches ambiguity before it turns into a bad shortlist.
2. Build an evidence-first scorecard
Create a 100-point candidate scorecard for this role. Every criterion must be observable in a public profile or resume. Separate must-haves from nice-to-haves and explain what evidence earns full, partial or zero points. Role: [paste role]
If a criterion cannot be observed, it should not silently drive ranking.
3. Expand titles without losing relevance
Starting from [title], generate direct equivalents, adjacent titles and titles that look similar but should be excluded. Explain the distinction in one sentence per group. Market: [industry and location]
The exclusion group matters as much as the expansion group.
4. Map the companies where talent is likely to sit
Map the types of companies most likely to employ strong [role] talent in [location]. Group them by industry, business model and company stage. Explain why each group is relevant and what would make it a false positive.
This replaces random company lists with a defensible talent-market hypothesis.
5. Reverse-map a client market
I recruit [role]. Describe how to identify US companies that already employ this role and are likely to hire it again. Return the buyer titles to target at startups, mid-market companies and enterprises.
Recruiters can use the talent market itself as a signal for future client demand.
6. Turn job openings into recruitment leads
Given these live job openings, rank the companies by likelihood to use an external recruiter. Use hiring volume, role difficulty, urgency signals, team growth and buyer accessibility. Jobs: [paste links or details]
A job post is a signal, not proof of agency demand. The ranking should explain the gap.
7. Find the actual hiring buyer
For a company hiring [role], list the most likely buyer titles in priority order. Adjust for company size and distinguish the economic buyer, hiring manager and internal recruiting owner.
This prevents sending the same generic pitch to HR, a functional leader and a founder.
8. Write outreach from evidence
Draft a short first message to [person] at [company]. Use only these verified facts: [facts]. Connect one fact to one recruiting problem. Ask one low-friction question. Do not use hype, fake familiarity or generic compliments.
The model should not invent personalization to make a message sound warmer.
9. Audit a shortlist for hidden assumptions
Review this shortlist against the role scorecard. Identify unsupported assumptions, duplicated profiles, missing talent pools and candidates who look relevant only because of title matching. Shortlist: [paste list]
Use the model as a critic after using it as a generator.
10. Build a five-day desk plan
Build a five-day operating plan for this recruitment desk. Include candidate sourcing, client mapping, decision-maker research, outreach and follow-up. Prioritize actions by expected placements or new business, not by activity volume. Desk: [describe niche]
The output should optimize for outcomes, not a longer task list.
One operating rule
Keep the raw brief, criteria, evidence and final decision connected. When those pieces live in separate tools, auditability disappears quickly.
I built Instalent to run candidate sourcing, client mapping, decision-maker research, scoring and outreach from one real brief. It has a free 7-day trial. No call is required.
Founder disclosure: I am the founder of Instalent. The prompts are useful without the product, and the GitHub playbook is public.
Top comments (0)