Training a Head of HR on AI Tools: Recruiting + Policy

When the head of HR at a 320-person services company told me she'd "stay out of the AI thing because of compliance," her team was already using ChatGPT for screening — just unofficially. That's the shape of HR-AI today, everywhere. The training job is to bring it into the light.

Why HR is the highest-stakes AI training in the company

Every other function deals with AI as a productivity question. HR deals with AI as a legal-and-trust question. That's why head-of-HR training has to start in a different place: not "what can AI do for us" but "what is our team already doing, and how do we make it safe."

The EU AI Act (with fines up to €35M or 7% of global turnover for high-risk misuse) treats hiring and people decisions as high-risk by default. Shadow AI in HR isn't just a policy gap — it's a liability waiting to materialize. Replika's €5M Italy fine, OpenAI's €15M Italy fine, and Clearview's €30.5M Dutch fine all came on training-data and privacy grounds. None of them were HR cases, but the regulatory direction is clear.

Definition: Augment, don't replace — AI drafts the rubric, the human makes the hire/fire decision. AI never owns a people decision.

The arc that works runs over six weeks: shadow-AI policy + safe-tier setup (week 1), recruiting workflows (weeks 2-3), policy drafting and Q&A (week 4), performance and L&D (week 5-6).

Where does the head of HR start?

Not with recruiting. With the policy that doesn't exist yet.

Step 1: the shadow-AI audit

Sit shoulder-to-shoulder with the head of HR and one HR-ops analyst. Pull (anonymized) examples of how AI is currently being used in the company — IT logs, helpdesk tickets, candid manager interviews. Use this prompt:

You are an HR policy analyst at a [INDUSTRY] company with [X] FTEs.
Below is a description of how employees are currently using AI tools.
(1) Identify uses that create regulatory risk (hiring, performance,
PII handling) and explain why for each. (2) Identify uses that are
low-risk and should be encouraged with light guidance. (3) Draft
a 1-page policy outline distinguishing the two categories with
clear examples. (4) Flag 5 questions the head of HR should validate
with legal counsel before publication. Do not invent practices
not in the source.

The head of HR sees the actual exposure surface. That conversation usually goes from "we don't have an AI problem" to "we have a written policy by Friday" inside one session.

Step 2: enterprise-tier setup

Most SMBs don't realize the difference between consumer AI tools (data trains the model) and enterprise tiers (no training, audit logs, SSO). 46% of employees admit uploading confidential data to public AI; the head of HR owns the migration to a safe tier. (See disclosure note.)

Tool tip (Course for Business): In our 6-week program the HR track always opens with shadow-AI containment because it's the only function where mis-sequencing creates legal risk. We pair the head of HR with one HR-ops AI Champion (1:15-20) who owns the policy library and approved-tools registry after the workshop. Augment, don't replace is a non-negotiable framing in HR — the head of HR signs off on every people decision regardless of what the model suggests. → https://course.aiadvisoryboard.me/business

Weeks 2-3: recruiting workflows

Recruiting is where the time-savings show up first and most dramatically. A translation-services company famously moved candidate screening from 3 hours to 3 minutes per role with structured AI rubrics — an 83% intake-efficiency gain. The mechanism is not "let AI pick candidates." It's:

1. Head of HR + hiring manager build a structured rubric (skills, experience, must-haves, nice-to-haves)
2. AI evaluates each resume against the rubric and outputs a structured score with reasoning
3. Recruiter reviews the AI output, makes the human shortlist call
4. Every rejected candidate gets a human-reviewed reason, not a black-box "no"

This is the intercom-fin pattern: AI-first with mandatory human escalation. Klarna walked back its full-AI customer service because escalation was thin; HR can't afford that mistake on hiring decisions.

Week 4: policy drafting and HR Q&A

Most heads of HR write 6-10 policies a year (remote work, expense, PTO, IT, AI itself, performance, conflict-of-interest, parental leave, return-to-office, security). Each one used to take 4-8 hours. AI compresses to 1-2 hours by drafting from existing policies in similar-sized companies, then having the head of HR shape and the lawyer review.

Workflow:

1. Paste the policy brief and 1-2 reference policies from sister companies
2. Ask the model to draft, flag legal-review points, propose a FAQ for managers
3. Head of HR edits, sends to counsel, publishes

The win is consistency — every policy from the same template, same voice, same FAQ structure.

Weeks 5-6: performance and L&D

Performance review prep, calibration meetings, development plans, training material — all draft AI-first from existing data. The head of HR becomes the first executive in the company with a fully-drafted package each cycle, instead of the one chasing managers for inputs.

Team scan (what AI champions report after week 1)

A typical 30-500-employee HR team after week 1:

• Adoption: Head of HR + 1 HR-ops Champion daily; 2-4 recruiters experimenting
• Use case #1: shadow-AI audit + policy draft — closes the biggest exposure
• Use case #2: structured candidate screening rubric — saves 2-3 hours/role
• Use case #3: policy draft generation — 1-2 hours instead of 4-8
• Use case #4: manager FAQ generation per policy
• Use case #5: performance-review prep packs
• Friction: "the model wrote bias into the rubric" — fixed with explicit must-have/must-not-have guardrails
• Risk flag: PII upload to wrong tier — addressed in week 1 policy
• Saved time: typically 6-8 hours/week per head of HR once Champion is up
• Honest miss: the policy needs lawyer review; budget for that, don't skip

Tool tip (Course for Business): Heads of HR in our 6-week program finish with: AI-use policy, approved-tools registry, structured screening rubric, policy-draft generator, manager FAQ template, performance-prep workflow. The weekly Shoulder-to-Shoulder review with the HR-ops Champion is what keeps the policy library current as new AI tools appear monthly. Augment, don't replace is reinforced every session. → https://course.aiadvisoryboard.me/business

Micro-case (what changes after 7-14 days)

A typical 240-FTE professional-services company trains its head of HR and senior HR-ops manager together in week 1. By day 7, an AI-use policy is published, three vendor consumer-tier accounts are migrated to enterprise-tier with audit logs, and the recruiter team is trained on a structured screening rubric. By day 14, candidate screening for two open roles drops from ~3 hours per role to under 30 minutes including human review, and the head of HR has cleared a backlog of three policies that had been "almost done" for six months. Most of the saved time gets reinvested in manager coaching and a long-overdue compensation-band review.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30-500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

FAQ

Should the head of HR learn to write prompts? Lightly. The head of HR writes prompts to draft policies, rubrics, and review packs. The HR-ops Champion writes prompts that build templates the team reuses. Different jobs.

What about AI bias in hiring? Real risk, not hypothetical. Mitigations: never let AI make the final cut, always require structured rubrics with documented must-haves, log every decision for audit, sample-review the AI's rejections monthly. The EU AI Act treats hiring as high-risk; assume regulator scrutiny.

Will AI replace recruiters? No, but it changes recruiter work. The realistic pattern is fewer-but-more-senior recruiters who own the rubric, vendor relationships, and candidate experience — while AI compresses screening volume. The 3h→3min translation-services example is screening, not full hiring.

How is this different from a generic "AI for HR" course? Generic courses lead with recruiting and skip policy. This leads with policy because the legal exposure compounds while the team waits. BCG found programs under ~5 hours produce no behavior change, but in HR the bigger risk is not skill — it's order.

What if the head of HR is non-technical? Most are, and that's fine. Every workflow is paste-and-prompt. The job is judgment, escalation discipline, and audit trail — not tech.

The takeaway

Heads of HR can't sit out the AI conversation. The shadow-AI is already happening in your company; the only choice is whether HR leads it or audits it. Train the head of HR on policy and approved-tools first, recruiting second, performance third. Pair them with one HR-ops Champion, run six Shoulder-to-Shoulder weeks, and HR becomes the function that protects the company while quietly multiplying its own throughput.

Next step: pull the IT log of AI-tool usage in your company and book the 90 minutes.

If you want every employee — including your head of HR — to ship their first AI automation in five days, book a 30-min call and we'll map your HR team's first week. → https://course.aiadvisoryboard.me/business