Recruiting is one of the most admin-heavy functions in any organization. Writing job descriptions, screening 200 applicants for 10 relevant ones, scheduling interviews, preparing structured questions, and keeping candidates warm through a long process — each step consumes hours that recruiters should be spending on the relationships and judgment calls that actually determine hire quality. AI does not evaluate candidates for you, but it compresses every administrative step in the funnel dramatically. Here is how to build the workflow from job requisition to offer letter.
Most job descriptions are written by copying last year's version and updating the team name. The result is generic postings that attract generic applicants. AI can help you write JDs that are specific, honest, and calibrated to the candidates you actually want to hire — not the idealized unicorn that nobody can find.
Write a job description for this role. Make it specific and honest — not generic. Role: [title] Team: [team context and size] What this person will actually do day-to-day: [list the real tasks] What success looks like in 90 days: [specific outcomes] Must-have skills: [hard requirements only — keep this short] Nice-to-have skills: [things that matter but won't disqualify] What we're NOT looking for: [helps filter mismatched applications] Salary range: [be transparent if possible] Work model: [remote / hybrid / in-office] Write in plain English. Avoid jargon. Keep the requirements section honest — every extra requirement reduces the qualified applicant pool by 20–30%.
Pay particular attention to the "must-have" list. Research consistently shows that requiring credentials or years of experience beyond what the role actually needs disproportionately filters out qualified candidates from non-traditional backgrounds. AI can flag when your requirements list is longer than the role warrants — just ask: "Which of these requirements would actually disqualify a high performer if missing?"
A posting for a mid-level role routinely generates 150–400 applications. Reading every resume carefully is not realistic, so most recruiters develop shortcuts — often unconscious ones — that miss strong candidates. AI can do a consistent first-pass screen based on explicitly defined criteria, which is both faster and more equitable than informal skimming.
I'm screening candidates for [role]. Here are my evaluation criteria: MUST HAVE (disqualifying if absent): - [criterion 1] - [criterion 2] STRONG SIGNAL (weight heavily): - [criterion 3] - [criterion 4] IGNORE: - Years of experience at big-name companies (not a proxy for skill) - Degree from specific schools (we hire on demonstrated ability) For each candidate below, output: - Meets must-haves: Yes / No / Unclear - Strong signals present: [list what you found] - Suggested action: Phone screen / Skip / Borderline — needs human review - One sentence of reasoning Candidate profile: [PASTE RESUME TEXT]
Important caveat: use AI screening to build a longer list for human review, not to make final skip decisions autonomously. The prompt above is designed to surface "Borderline — needs human review" for ambiguous cases rather than silently discarding them. Human bias is a real risk in recruiting; AI introduces different risks (including training data biases) — the goal is to make the screening step faster and more consistent, not to remove human judgment from candidate evaluation.
Unstructured interviews are notoriously unreliable predictors of job performance — and also the most common format. Structured interviews with consistent, role-specific questions significantly improve predictive validity. The problem is that writing good structured questions is time-consuming. AI can generate a complete interview guide from your job description in under two minutes:
Create a structured interview guide for this role: Role: [title] Key competencies to assess: [list 4–5 things that predict success in this role] Stage: [Phone screen / Technical / Hiring manager / Panel] For each competency, write: 1. One behavioral question (past behavior: "Tell me about a time...") 2. One situational question (hypothetical: "How would you handle...") 3. What a strong answer looks like (2–3 sentences — the evaluation rubric) 4. One follow-up probe if the initial answer is vague Also include: - 2 questions the candidate should ask us (signals of a serious candidate) - Red flags to watch for in this type of role - Scoring guide: 1–4 scale with descriptors for each level
The resulting guide makes your interviews more consistent across interviewers, makes calibration conversations easier ("they scored 2 on the collaboration competency because..."), and gives you defensible documentation if a hiring decision is ever questioned.
Cold outreach to passive candidates has a response rate problem: most messages are obviously templated, and experienced candidates have seen them all. AI can help you write personalized first-contact messages that reference something specific about the candidate — their work, their publications, their career trajectory — making them far more likely to respond.
Write a personalized recruiting outreach message for this candidate. Role I'm hiring for: [title + 1 sentence on the team] What makes this role compelling: [be specific — growth? mission? tech stack?] Candidate background (from LinkedIn or portfolio): [paste relevant details: current role, notable projects, background] Constraints: - Under 120 words - Reference one specific thing from their background - Lead with what's in it for them, not what we need - End with a low-commitment ask (15-min call, not "apply now") - No corporate jargon, no "I came across your profile"
Use Make.com to automate parts of this workflow: pull candidate data from LinkedIn exports or your ATS, send to Claude with the prompt above, and populate a message template. You still send messages individually, but the drafting step is automated. For sourcing at scale, this can save 3–5 hours per week on outreach drafting alone.
Candidates drop out of processes most often because of silence — not because of competing offers. A consistent communication cadence, even when there is no new information, dramatically reduces ghosting and improves offer acceptance rates. AI can draft status update emails for every stage of your process:
Write a candidate update email for this stage: Stage: [After phone screen / After panel interview / While references are being checked / etc.] Timeline context: [what's happening next and when] Tone: Warm, professional, honest — never corporate-robotic Keep it under 100 words. Be specific about what happens next and give an exact timeline if you have one. If you don't have a firm date, say that directly rather than giving a vague "soon."
Templates for each stage can be built into your ATS or triggered automatically via Make.com when a candidate's stage changes — creating a communication cadence that runs without recruiter intervention for the routine updates.
Every candidate who goes through your process is a potential future employee, customer, or referral source. Rejection emails that feel automated and cold damage your employer brand; offer letters full of legalese and ambiguous compensation tables leave candidates confused. AI produces better first drafts of both:
For rejections: provide the candidate's stage and a brief reason (even if you share only a general version), then prompt for a warm, specific, 3-sentence message that thanks them, gives the honest reason in a way that is useful feedback, and leaves the door open. For offer letters: provide all the compensation components and prompt for a plain-English summary alongside the formal letter — candidates should not have to decode their total comp package.
If your team uses Notion for HR documentation, see our guide on AI tools for HR teams for a broader look at how AI fits into the full people-operations workflow, from job posting through performance reviews.
AI cannot assess cultural fit, read the room in an interview, or sense the interpersonal dynamics that determine whether a candidate will thrive on a particular team. It cannot make the judgment call to pass on a technically strong candidate because they will clash with a key stakeholder, or to take a chance on an unconventional hire because you saw something in them the resume did not capture. These judgment calls are the highest-value work in recruiting, and they require human presence and experience.
What AI removes is the administrative drag that prevents recruiters from doing that work well. When you spend four hours writing JDs and screening resumes that could be compressed into forty-five minutes, you have three fewer hours for the relationship-building and evaluation work that nobody else can do. AI shifts that ratio — and in recruiting, that shift compounds directly into better hires and faster pipelines.
💡 Ready to build your recruiting AI stack? Browse the full toolkit →
Practical prompts and automation ideas — no fluff.