As of 13 August 2026, AI can only partly manage your recruitment pipeline.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open the approved job description and your recruitment process document, then confirm the essential criteria, desirable criteria, stages, owners and decision rules.
   2. Export or copy the current pipeline from your ATS or spreadsheet, remove unnecessary personal data, and keep only candidate identifiers, application evidence, stage, dates, notes and next action.
   3. Paste the job description, pipeline data and process information into the prompt, replacing each bracketed section with your material.
   4. Ask the model to produce the candidate table, overdue actions, draft messages, interview questions and human decision checklist.
   5. Compare every candidate summary and overdue action against the source application and ATS record, correcting any invented evidence, dates or status.
   6. Ask a recruitment or HR colleague to check the proposed criteria, questions, shortlist handling and messages for consistency, accessibility, bias and data protection risks.
   7. Enter only approved stage changes into the ATS or pipeline, send candidate messages after a human approves them, and record the human decision and reason.

   Prompt

   Act as a recruitment operations assistant for a UK employer. Using only the information I provide, turn this recruitment pipeline into a clear working plan.
   
   Role and hiring criteria:
   [Paste the approved job description, essential criteria, desirable criteria, salary information and location or working pattern]
   
   Current pipeline data:
   [Paste candidate identifiers, application status, dates, interview status, notes and next actions. Remove unnecessary personal data]
   
   Recruitment process:
   [Paste the stages, owners, service targets, interview process and decision rules]
   
   Produce:
   1. A table of every candidate by current stage, stated evidence against each essential criterion, missing information, owner and next action.
   2. A list of overdue or blocked actions based only on the dates and targets supplied.
   3. Draft neutral, accessible messages for candidates who need an update, interview invitation or rejection, with placeholders where a human decision or reason is missing.
   4. Interview questions mapped to the essential criteria, avoiding questions about protected characteristics or unrelated personal circumstances.
   5. A short human decision checklist for each stage.
   
   Do not rank or reject candidates automatically. Do not infer skills, health, disability, family circumstances, nationality, age, religion, sex, sexual orientation or any other protected characteristic. Do not invent candidate evidence, reasons, dates or legal conclusions. Mark uncertainty clearly. Flag inconsistent criteria, unexplained gaps, possible bias, duplicate records, missing consent or data that should not be retained. Treat all final shortlisting, rejection, adjustments, data protection and employment decisions as human decisions. Use British English and return the output in clearly labelled tables and bullet points.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.