Skip to content
Two broad terracotta and slate-blue pencil fields overlap above a sage lower edge on warm ivory paper.

Role and function playbooksArticle

How should a recruiter use AI for drafting and interview preparation?

Prepare job adverts and interview questions from agreed role facts, catch unsupported requirements and authority changes, and review the complete pack before release.

Jump to a section

Use AI to prepare a first draft from agreed job facts, then check what the draft changes before anyone uses it. For a recruiter working with a hiring manager, the useful output is a job advert or proposed interview guide whose requirements can be traced to the actual role. Fluency is only one part of that review.

Keep authority over the role and the assessment with the people responsible for them. An assistant can suggest wording or questions. It cannot settle an unresolved salary, add a qualification or turn a question list into an approved way to judge applicants. This article covers employer preparation, before candidate assessment. The wider roles and industries guide explains how to bound an AI trial around a particular piece of work.

Agree the job facts before asking for a draft

A job title is a thin brief. Ask the hiring manager what the person will actually do, which decisions they can make and what evidence supports each requirement. Bring the current approved role description and employment terms to that conversation. Resolve differences between those documents and the manager's expectations before generating candidate-facing material.

The U.S. Office of Personnel Management's job analysis guidance connects the tasks in a job with the competencies needed to perform them. That is a useful preparation discipline. It does not mean that pasting a job description into an assistant completes a job analysis.

In a September 2026 Reddit discussion, a Reddit user described incomplete job descriptions and hiring managers changing the profile after seeing candidates. The commenter also distinguished matching credentials from deciding whether experience transfers to the actual team and work. This is an observation in a discussion, with no independently verified employer or measure of how often it happens. It exposes a problem worth checking before drafting: the people commissioning the hire may not yet agree on the job.

Suppose an appliance retailer is hiring a customer-support supervisor. Recruiter Jamie Lee meets hiring manager Alex Morgan to prepare the advert and interview material. The approved role includes prioritizing the support queue, coaching colleagues and escalating refund exceptions to the authorized manager. It does not give the supervisor authority to waive the refund policy.

Jamie keeps a short preparation record:

Part of the roleSource or owner to checkWhat the draft must preserve
Queue priorities, coaching and refund escalation
Current role description, confirmed with Alex
Describe the actual work and the supervisor's decision limits
Essential skills and qualifications
Agreed role requirements and the recruitment owner
Separate required capabilities from preferences; add no new credential
Pay, location, hours and benefits
Approved employment terms and their owner
Use confirmed terms; flag missing information instead of guessing
Interview purpose and assessment method
Existing approved hiring plan and assessment owner
Propose material within that plan; do not invent new scoring criteria

If Alex now expects the supervisor to approve exceptional refunds, Jamie has found a role decision to resolve. Rewording “escalate” as “resolve independently” would hide it. Update the approved role and its authority through the appropriate process before asking an assistant to describe the change.

Draft the advert without adding requirements

Give the assistant the agreed material in an approved workplace tool, using only information permitted for that purpose. This preparation task does not need a candidate's CV, interview transcript or personal case history.

A bounded instruction for Jamie's case could be:

Draft a customer-support supervisor advert using the approved role description and employment terms below. Preserve the distinction between handling ordinary refunds and escalating exceptions. For each requirement, identify the supporting source section. List missing or conflicting information separately. Do not invent pay, qualifications, hours, benefits or decision authority.

Source references make checking more convenient; Jamie still has to open the source and verify the match. An assistant can attach a convincing reference to a sentence the document does not support.

Microsoft's recruiting scenario illustrates drafting a job description and brainstorming interview questions. Treat it as a demonstration of possible tasks. It does not establish that a particular draft is accurate or that using it improves hiring outcomes.

Read the advert in two passes. First check the facts against the approved material. Then read it as a prospective applicant would, asking whether someone can understand the work and the actual requirements.

  • Find additions. Look for a degree, years of experience, software expertise or availability requirement that nobody approved.
  • Check the level. Words such as “own,” “lead” and “authorize” can quietly expand responsibility. Compare them with the role's real decisions.
  • Inspect exclusions and promises. Challenge vague descriptions of an ideal personality, unexplained jargon and unsupported claims about flexibility or progression.
  • Keep gaps visible. A missing pay range belongs with the person who approves it. It should not become a plausible number in the advert.

The UK's Responsible AI in Recruitment guidance identifies risks from vague or dissuasive job-description language and discusses effective oversight. Its March 2024 assurance guidance supports examining the output and its effects; asking another assistant to declare wording “unbiased” does not settle that review. Use the organization's recruitment, privacy and legal owners for the actual use and jurisdiction.

Trace each proposed interview question to the work

Generate a small set of proposals that the reviewer can inspect properly. Ask for the job requirement each question is intended to explore, the source supporting it and any assumption needed to answer. Keep missing information separate from the question itself.

For Jamie's supervisor role, a draft might ask how a candidate would exceed the refund limit to retain an important customer. That sounds like a customer-service question, but it rewards authority the supervisor does not have. The error concerns what the job permits, even if the sentence is clear and the answer would make an interesting conversation.

A customer-service question can change the job

Approved duty

Escalate refund exceptions

The supervisor handles ordinary refunds under the policy and sends exceptions to the authorized manager.

Unsupported proposal

Exceed the limit to retain a customer

How would you approve a refund above your limit to keep an important customer?

Revised proposal

Explain the escalation

A customer requests a refund above your limit. What would you explain to the customer, and what information would you pass to the authorized manager?

Example. The first proposal assumes authority the role does not have. The revision preserves the limit but still needs review within the approved assessment method.

The repaired question gives Alex a proposal to review. Alex still needs to check whether it fits the agreed assessment purpose, whether applicants receive enough context and whether the expected response is appropriate for the role. A more accurate question is not automatically a validated assessment.

A different error is subtler: the assistant may label a question “leadership” when it only asks someone to describe a familiar process. Check the evidence a response could actually supply. If the hiring plan needs evidence of coaching, an unrelated policy-recall question will not provide it merely because the generated label says so.

OPM's description of structured interviews includes predetermined questions and common rating standards. A generated list supplies neither a complete assessment method nor permission to change one. Keep the established questions, approved follow-up approach and evaluation requirements under the assessment owner's control. If new criteria are needed, pause preparation for the appropriate review rather than filling the gap with a generated scorecard.

When the actual vacancy is an AI adoption leader, the separate guide to interviewing an adoption leader covers the practical exercise and evidence to examine. Use this preparation check before those materials reach the interviewers.

Review the complete pack before releasing it

The advert and interview guide need to describe the same job. A careful check of individual sentences can miss a disagreement between documents. Jamie compares the complete pack with Alex before handing it to the person authorized to release it.

One employer makes that review explicit in its public account. TEKenable's AI-use guidelines say its team reviews AI-assisted job descriptions for accuracy and role alignment, and recruiters and hiring managers approve proposed questions before use. That is the employer's stated process, rather than independent evidence of its results. The useful feature is the visible handoff from drafting to accountable review.

For the retailer's pack, work through the checks in order:

  1. Reconcile the sources. Confirm that the role description, employment terms and hiring plan are the current agreed versions. Stop if they conflict.
  2. Resolve the edits. Record what changed in the advert and questions, including anything removed because it lacked support. Send unresolved role decisions back to their owner.
  3. Read across documents. Check that the advert does not promise authority or conditions the interview guide contradicts. Keep the approved assessment method intact.
  4. Release one agreed version. Have the authorized owner accept the pack. Give interviewers that version and a route to raise a concern or request an adjustment through the established process.

If the role changes later, revisit the affected materials together. Quietly changing one interviewer's copy creates a different problem from correcting an unused draft. Ask the assessment owner how a proposed change should be handled before using it with applicants.

Count the work of getting to an acceptable pack

A fast first draft can still create a long review. For a bounded trial, compare the whole preparation task with the current method: gathering facts, drafting, checking, correcting and obtaining approval. Record who did that work, including time taken from the hiring manager.

Inspect the outcome as well. Did the pack preserve the agreed requirements? Were missing facts caught before release? Did Alex have to correct an authority change that would have affected the questions? Keep those findings beside the effort record. Do not turn fewer writing minutes into a claim of better hires.

If source checking takes more effort than drafting the pack directly, change the task or stop using the assistant for that part. A useful result may be a clearer source record or one good question proposal. It need not justify extending the tool into candidate screening. For the wider HR remit, review internal service quality separately from organization-wide enablement. A checked recruitment pack supplies evidence about this task, rather than another department's readiness.

Questions about AI-assisted recruitment preparation

Can a recruiter use AI to write a job description?

A recruiter can use an approved tool to draft wording from agreed role facts and permitted source material. The recruitment owner must still check requirements, employment terms, clarity and the role's decision limits before release. If the manager cannot confirm what the job includes, resolve that first. The preparation record shows the inputs to establish.

Do interview preparation prompts need a candidate's CV?

Preparing the core questions for a role can begin with the approved job requirements and assessment plan, without a candidate's CV. Adding personal information or tailoring questions to individuals changes the task and may affect data handling and consistency. Refer that proposed use to the appropriate recruitment and privacy owners instead of uploading files because a tool accepts them.

Are AI-generated interview questions a structured interview?

A generated question list alone is insufficient. A structured interview requires an agreed method, including consistent questions and evaluation standards. Have the assessment owner review how proposed questions connect to job requirements and fit that method. Do not adopt a generated scoring scale as evidence that the assessment is valid or fair. The question-repair example concerns preparation, before those judgments.

Can AI fill in a missing salary range?

Do not let an assistant invent a salary range for an advert. Ask the compensation owner for the approved terms and resolve the evidence gap before release. For an AI adoption manager vacancy specifically, match pay comparisons to the role's level and responsibility. That article does not establish a salary for a customer-support supervisor or another unrelated job.

How should a recruiter decide whether AI preparation is worth continuing?

Compare the complete preparation effort with the existing method, including the hiring manager's review and all corrections. Inspect whether the final materials preserve the agreed job facts and assessment boundaries. Continue only where that combination is useful; a faster draft with unsupported requirements is insufficient. Keep candidate screening, ranking and hiring decisions outside this writing trial.

Updated

aiready

A home for your company’s AI community.

Share what works and help each other put AI into practice.

  • Real use cases

  • Practical guides

  • Company policies

  • Shared experience

Explore aiready
Explore the blog