Hire an AI applicant evidence and hiring reply drafts

Everything it can hand back

Your task sets the actual outputs. The specialist picks the methods that cover them.

• A job post and a scoring guide, made once and reused
• A running ranking with the quoted evidence behind each score
• Draft invites, replies and holding notes, all marked not sent
• A call on whether to widen the search, and where to look
• A one-page update each Friday

How it does the job well

• Base people decisions on job-related evidence, applied the same way to everyone.
• Write in plain, respectful language the person affected could read.
• Keep personal data to what the task needs, and handle it privately.
• Check local employment law and company policy before advising on hiring, pay or exits.

Mistakes it avoids

• Scoring on anything that is not in the scoring guide.
• Treating a low score as a final no. It is a prompt for a person to look.
• Writing a rejection that gives a reason you cannot back with the scoring guide.
• Running again on the whole pile when only a few applicants are new.
• Sharing one applicant's details with another, or in the Friday update beyond what the hiring manager needs.
• Biased wording in job ads, reviews or interview questions.

What a good result looks like

An example, not your task: You have an open role and applications keep arriving in batches.

• post.md. The job post for the role, made once and reused.
• scoring-guide.md. The must-haves and nice-to-haves. Only these are scored. You approve it before anyone is scored.
• ranking.md. A running ranking of applicants, with the quoted resume line behind each score.
• seen.md. The applicants already scored, by ID, so no one is scored twice.
• drafts/. Draft invites, replies and holding notes, each marked not sent.
• Path taken. Whether to go on to replies or widen the search, and why.

How the work is checked

• Every score is against the scoring guide and has a quoted line from the resume behind it.
• Nothing about name, age, gender, photo, address, school, employment gaps or how the resume looks affected a score.
• No message was sent, no interview booked and no one rejected without a clear yes.
• No winner was named. The person makes the hiring call.
• States which checks passed, failed or could not run, and names missing inputs.

These checks have not been run against a model yet. Your AI uses them to check its own work.

Capabilities

• Get the job post and scoring guide. Builds them once for the role and shows them to you before anyone is scored. Only what is in the guide counts.
• Find the new applicants. Compares each batch with the people already scored, so no one is scored twice.
• Screen the batch. Scores each new applicant against the guide and adds them to the ranking, with the quoted line behind each score.
• Choose the path. With enough strong fits, it goes on to replies. With too few, it suggests where else to look and flags a must-have that may be set too high.
• Draft the replies. Writes an invite, a clear reply or a holding note for each person. Every draft is marked not sent.
• Compare finalists. After interviews, lines up your notes against the guide. It does not name a winner.
• Report on Friday. Counts people by stage, shows what is stuck and what to do next week.

How the role works

1. Checks your task, context and available tools.
2. Picks the methods needed and prepares a work plan.
3. Carries out supported tasks, with each method's limits and review gates.
4. Checks the real deliverables and corrects supported faults.
5. Hands back files, evidence and unresolved gaps.

Your specialist's working setup

Download the base specialist without a task form or account. README.md and SETUP.md explain how to use the folder. SOUL.md defines its scope. INTAKE.md asks missing questions inside your AI. PLAYBOOK.md, SKILLS.md, BEST_PRACTICES.md, CHECKS.md and HANDOFF.md guide the work and its checked deliverables.

Complete associated skills, supporting files and licences are included. Choose only the methods needed for the actual task. Keep context, working files and accepted feedback in your AI project.

Use this specialist through MCP or the API

Call prepare_specialist with role_id hiring-pipeline-runner only. No task is required to load it. Your AI asks what job you need done, what you want back and which relevant context it can use.

For task-led hiring, call find_specialist, explain the suggested specialists and let the user accept or change the team.

Complete ZIP: Download base specialist. API: GET /api/v1/specialists/hiring-pipeline-runner/pack. The host AI supplies its own model, tools and execution; a pack does not connect accounts or run a task.

Inputs and helpful context

The specialist starts from the task brief. Only active methods require their inputs.

For the primary method, when used

• A short brief of the role and what the team needs
• Each new batch of resumes or application forms
• Your notes from each interview, after the interviews (optional)
• Your country, so the drafts and checks fit local rules

Helpful context (optional)

• The scoring guide you used for your last hire
• A list of the stages in your hiring process

Tools and access

• Claude Code: It can keep the hiring folder and the list of who was scored. You start each run, including the Friday update. It does not run on a timer.
• Gmail: Applications often arrive here. Export or attach the batch. Sending stays with you.
• Any good chat AI: This is a writing job. Claude, ChatGPT or Gemini all do it well, with nothing to install.
• Any good chat AI: A chat AI can plan and write all of it. It cannot run your ads or read your ad accounts.

Functions and deliverables

• Applicant evidence and hiring reply drafts: Review new applications against an approved job guide. Get cited scores, reply drafts and gaps for a person to assess.
  • A job post and a scoring guide, made once and reused
  • A running ranking with the quoted evidence behind each score
  • Draft invites, replies and holding notes, all marked not sent
  • A call on whether to widen the search, and where to look
  • A one-page update each Friday
• Hiring packet builder: Turn a hiring brief into a job post, scored interview guide, and offer letter template.
  • A job post
  • An interview guide with scoring rubric
  • An offer letter template
  • A draft DocuSign link if you ask for it
  Source: Anthropic.
• Applicant shortlist builder: Give it the job and the resumes. Get a ranked shortlist scored only on the job needs, with replies drafted.
  • A rubric if you lack one
  • A ranked shortlist with evidence
  • Fairness notes on what was ignored
  • Draft replies and interview invites
  Source: Anthropic.
• Candidate message drafter: Write the right email to a job candidate at any stage, from screening invite to offer or rejection, with one clear next step and a send time.
  • A subject line and email draft for that stage
  • One clear ask for the candidate
  • A suggested send time
  • A warmer and a more direct wording for the key lines
  • A list of any gaps you still need to fill
  Source: Carlo Pezzuto.
• Candidate sourcing plan writer: Share a role and candidate profile. Get ranked search channels, target companies, search terms and a four-week plan.
  • Ranked sourcing channels for the role
  • Target companies and communities
  • A week-by-week action plan
  • Notes on what to do first
  Source: Carlo Pezzuto.
• Job ad campaign planner: Plan a recruiting campaign for one role: channels and budget, job ad copy, a 4-week content calendar and an A/B test plan.
  • A channels and budget table
  • Job ad copy and three headline options
  • A four-week content calendar
  • A test plan with what to change and measure
  • A funnel report layout from views to hires
  Source: Tuan Duc Tran.
• Fair candidate comparer: Compare job candidates fairly. Paste interview transcripts and your scorecard. You get scored answers, evidence gaps and bias checks.
  • A scorecard filled in question by question, with quotes
  • A written screen evaluation grouped by skill
  • A side-by-side comparison of candidates with a confidence level
  • A list of evidence gaps, interviewer gaps and bias flags
  Source: Carlo Pezzuto.
• Hiring progress reporter: Turn your hiring pipeline numbers into a clear update for your hiring manager or leaders. Pick a weekly update, market report, talent map or look back.
  • A readable pipeline update
  • A market or talent map report
  • A simple chart page if you ask
  • Specific recommendations
  Source: Carlo Pezzuto.
• me.md builder: Answer a few questions, or let your AI use what it remembers. Get one private file that tells any AI who you are.
  • One me.md file with up to ten dated sections
  • A list of facts taken from memory, for you to check
  • A short note of what it removed for privacy
  • The two next steps, to save it and set it up

Tasks you can use it for

• You have an open role and applications keep arriving in batches.
• You want every applicant scored the same way, with the reason shown.
• You want a short update each Friday without building it by hand.

Limits to check before starting

• You plan to let AI screen applicants without a check first. Many places have rules for AI that screens or ranks job applicants. Check data protection and anti-discrimination law in your country before you use it.
• You need to decide who to hire or make an offer: that stays with you, with a qualified HR adviser or employment lawyer for the rules.
• You only want one job post and scoring guide before any resumes arrive: use the Hiring packet builder.
• You only have one small pile of resumes to rank once: use the Applicant shortlist builder.

Best way to use this role

Give it one clear outcome, the actual files and a definition of done. Keep the work in one AI project so it can reuse checked context and earlier files.

• Start a chat with the agent. In Claude Code, add it as a skill or save it in .claude/agents, then give it the role brief.
• Before the first batch, check whether AI screening of applicants is regulated where they live (for example the EU AI Act, or New York City's bias audit and notice rule). The agent asks you and records the answer. If you don't know, it stops until you have checked.
• Read the job post and scoring guide it makes, and change anything you disagree with before the first batch is scored.
• Before you paste resumes, check your AI tool's data terms and your local privacy rules, and take out names, photos and contact details where you can.
• Each time new resumes arrive, put them in the folder or attach them and ask it to run the new batch.
• The widen-the-search rule (fewer than three strong fits after two batches, or ten days) is a starting point. Set your own for the role.
• Read the ranking and the drafts. Review the strong-fit invites first, then the holds, then the replies to people who missed a must-have. Send, book or reject yourself, or say yes to one named action.
• After interviews, give it your notes and ask for the comparison. Run it again on Friday for the update.

Read it as data

Read the complete role setup as JSON. Connect your AI to the AI Blocks MCP library and read aiblocks://roles/hiring-pipeline-runner for its role instructions and pinned sources.

This is a prepared AI role. No task has run. Your AI carries out authorized work using its available tools and reports the actual deliverables and checks.