20 questions recruiters ask an AI assistant every day

Recruiters now spend most of the working day next to an assistant, and the questions they ask it have settled into a recognizable twenty. They cluster into four groups: sourcing (what am I looking for, and where), outreach (why is nobody replying), interviewing and screening (who is actually good), and strategy (why are we losing hires we should have made). This post is the index: the short answer to each question, then the link to the full playbook. Read together, the twenty are a fair portrait of what the job has become: less typing and searching, more judgment.

Sourcing: where is the rest of the market?

Most sourcing questions are one question in different clothes: my search is only showing me a sliver of the market, where is everyone else?

1. “Write me a Boolean string for this role.” The grammar takes five minutes: AND narrows, OR widens, NOT excludes, quotes lock phrases, parentheses group. The craft is the synonyms you feed it and the willingness to revise against the first two pages of results. The worked example, plus the LinkedIn quirks that silently break strings, is in Boolean search strings for recruiters.

2. “What other job titles should I be searching?” One title misses most of the market, because the person you need may carry any of a half-dozen labels. The repeatable method for mapping adjacent titles and skills, with a worked SRE example, is in sourcing with adjacent job titles and skills.

3. “Where do I find these people if not on LinkedIn?” In the communities where they do the work: GitHub, design forums, Slack groups, conference lineups. Candidates found there arrive with context attached, which is what makes the eventual outreach writable. The map by role, plus the etiquette, is in where to find candidates beyond LinkedIn.

4. “Turn these intake notes into a job description.” The draft takes seconds; the value is in the questions that surface what the hiring manager actually needs before anything gets drafted. Those questions, a JD skeleton, and how to use an AI draft well are in turning messy intake notes into a job description.

5. “Is this job ad too generic, or quietly biased?” Usually at least one of the two. The audit covers empty clichés, coded language, inflated requirements and missing salary information, with before and after rewrites, in is your job description too generic or biased.

6. “What is the difference between a data engineer and a data scientist?” Pipelines versus models, roughly, and the CVs look interchangeable until you know which signals to read. How to tell them apart at screening, and where analytics and ML engineers fit, is in the recruiter’s guide to data engineer vs data scientist.

7. “Which of these ten requirements are real must-haves?” Usually three or four. A practical test for a true must-have, the argument for capping the list at five, and the funnel math of shorter lists are in must-have vs nice-to-have skills.

Outreach: why is nobody replying?

Because the message reads like it was sent to two hundred people, mostly. The fixes are specific.

8. “Draft a cold message this person would actually answer.” Four parts: a subject line specific enough to earn the open, one genuine detail of the candidate’s work, the role in two sentences with salary and location in daylight, and an ask answerable in one line. Under 100 words, total. The anatomy, with a full worked InMail, is in cold outreach messages that get replies.

9. “No reply. What does the follow-up say?” Most replies come from the second or third touch, so the sequence deserves as much craft as the opener: spacing, a channel switch, new value in every message, and a graceful stop that does not burn the lead. The playbook is in follow-up messages candidates actually answer.

10. “How do I pitch a role that pays less than the candidate earns now?” With the number early and the honesty visible: state the range up front, explain equity in plain terms, sell flexibility and scope that are actually real, and recognize when the gap is too big to close. The script is in how to pitch a role with a lower salary.

Interviewing and screening: who is actually good?

The assistant supplies questions and structure here. The read on the person stays yours.

11. “Give me behavioral questions for pressure and conflict.” Six earn their place, and the probing matters more than the questions: almost anyone survives the first question, very few survive the third follow-up. Good-answer signals and red flags per question are in behavioral interview questions for pressure and conflict.

12. “How do I screen technical skills I do not have myself?” With questions where the shape of the answer reveals real depth, so a recruiter without an engineering background can score the response credibly. The bank is in technical screening questions that verify real skills.

13. “Which CV red flags actually matter?” Fewer than recruiters act on. Some patterns predict problems; others are noise that gets over-weighted, and almost every flag is better verified with a question than punished with a rejection. The split is in CV red flags worth checking, and ones to ignore.

14. “How do I screen 400 applications without turning into a form letter?” AI triage with human checkpoints: let ranking order the queue, keep a person on every consequential call, spot-check the rankings weekly, and make sure every applicant gets a real answer. The system is in screening high volumes without losing the human touch.

15. “How do I reject someone after the final round?” With a phone call, not a template. A final-round candidate invested weeks; what to say, feedback that is honest and legally safe, and a short note that keeps the door open are in rejecting a final-round candidate the right way.

Strategy: why are we losing hires we should have made?

The strategy questions arrive late in the day, usually after something went wrong. The answers point earlier in the process.

16. “What does a process that keeps candidates engaged look like?” Candidates quit silent pipelines, not slow ones. Stage SLAs, an update at every transition, self-scheduling, and a weekly review of stuck candidates do most of the work; the design is in a recruitment workflow that keeps candidates engaged.

17. “Where exactly are we losing people mid-process?” At four predictable points: post-apply silence, scheduling friction, too many rounds, and slow final decisions. Each leak has a fix and a metric to watch in stop losing candidates during the interview process.

18. “What should this role actually pay?” Numbers go stale; method does not. Triangulating salary surveys, live job ads and what candidates tell you into a range you can defend is covered in how to research market salaries for tech roles.

19. “My hiring manager wants ten years of experience in a five-year-old framework. Help.” Market evidence beats opinion: reframe at intake, run a pilot search and let the results argue for you, and escalate only with data in hand. The diplomacy scripts are in managing unrealistic hiring manager expectations.

20. “Why did they turn down our offer?” For one of five reasons that was visible weeks earlier: a faster competing process, late salary discovery, a counteroffer, a stricter office policy than implied, or an interview experience that quietly discounted the offer. The pre-closing routine that makes offers a formality is in why candidates turn down offers.

What does this list say about the job?

None of the twenty is “do my job for me”. Every one delegates preparation and keeps the decision, which is exactly the boundary an honest map of recruitment automation draws. The split holds across all four themes:

Theme	The assistant’s share	Your share
Sourcing	Expand strings, map adjacent titles, rank the pool	Which trade-offs the role can absorb
Outreach	Draft, translate, sequence, remind	The one personal detail, and every send
Interviewing	Question banks, scheduling, comparisons	The read on the person in the room
Strategy	Pipeline reports, salary legwork, stuck-candidate lists	The conversation that changes a hiring manager’s mind

That right-hand column is the job now. The left-hand column used to be most of the working week.

Where does Recruitifly fit?

Every answer above is, in one sense, generic: it applies to any team. The assistant you ask all day should not be, because there is a difference between “here is how follow-up sequences work” and “these six candidates have been waiting more than a week, here are the drafts”. That is the argument for the assistant living inside the ATS, where the data lives; the longer version is in what is an agentic ATS.

Recruitifly is built that way. One assistant, Fly, works across the whole platform: it searches your talent pool, scores and ranks candidates against a job, builds shortlists and side-by-side comparisons, drafts outreach and follow-ups in multiple languages, proposes interview slots and books them when you confirm, and assembles pipeline reports from your actual stages. Every write is propose-then-confirm: Fly prepares, you approve, and nothing reaches a candidate unsupervised. For hiring, we think that is the correct design rather than a limitation.

We are in private beta. If your day sounds like the twenty questions above, talk to us and bring a live role.