8 practical examples of AI in HR (with prompts)

AI is already showing up in everyday HR work, from drafting communications and summarizing feedback to analyzing workforce patterns and supporting routine processes. The challenge is knowing where it adds real value, where it creates unnecessary risk, and what should remain firmly in human hands.

The eight examples below show what practical AI use in HR can look like. Each one is tied to a specific task and includes a tool, framework, or prompt you can adapt to your own work.

Across all eight, the same principle applies: use AI to reduce repetitive work, organize information, and surface evidence, while keeping human judgment, accountability, and appropriate review in place. The closer a use case gets to a consequential employment decision, the stronger the safeguards need to be.

None of this is future tense. In organizations that have implemented AI, HR professionals report using it frequently: 26% weekly, 20% daily, and 9% several times a day.

8 examples of AI in HR

AI can support a wide range of HR work, but the right use depends on the task, the data involved, and the level of judgment required. The examples below range from lower-risk drafting and administrative support to higher-risk uses that require stronger controls, validation, and human oversight.

Here are eight practical examples of AI in HR:

1. Organize high-volume applicant information against predefined job criteria
2. Draft performance review summaries from approved, anonymized feedback
3. Prepare for a pay-equity review
4. Review anonymized attrition and engagement patterns for workforce trends
5. Personalize onboarding communications and reduce routine administrative work
6. Draft job descriptions from approved role information
7. Answer routine employee questions from approved HR sources
8. Organize interview and assessment evidence for qualified human review

1. Organize high-volume applicant information against predefined job criteria

Instead of skimming 200 résumés from scratch, start with the job-related criteria your hiring team has already defined. Use AI to extract and organize relevant information from applications against those criteria, making the evidence easier for a recruiter to review consistently.

Review the AI-organized information against the original applications before deciding who advances. The tool should surface evidence, missing information, or areas that require review, not rank candidates or recommend who should move forward. That second check matters because automated screening at scale can multiply historical bias as easily as it multiplies speed.

Once your criteria are defined, use a prompt like this to organize the evidence consistently:

Act as a recruiting support assistant. Review the applicant information below against only the predefined job criteria provided.

For each criterion:
- identify the evidence explicitly stated in the application
- quote or summarize the relevant experience without adding assumptions
- mark the criterion "not found" if the application does not provide enough information

Do not score, rank, compare, or recommend candidates. Do not infer qualifications, characteristics, or experience that are not explicitly stated.

Job criteria:
[PASTE PREDEFINED JOB-RELATED CRITERIA]

Applicant information:
[PASTE APPROVED, APPROPRIATELY HANDLED APPLICANT INFORMATION]

If you use an automated employment decision tool in New York City, Local Law 144 sets three clear obligations:

Federal antidiscrimination law still applies when an employer uses an automated system. The EEOC notes that the familiar four-fifths selection-rate test can flag potential adverse impact, but passing it does not automatically establish that a process is lawful.

Jenelle Buatti, an instructor in Ziplines' AI for HR Professionals course, puts the operating principle plainly:

"We cannot outsource accountability."

Best for high-volume requisitions where sheer volume is the bottleneck. Start with a limited pilot, and do not proceed if you cannot validate the tool, review its outputs, protect candidate data, and meet the requirements that apply where you hire.

[[cta-big]]

2. Draft performance review summaries from approved, anonymized feedback

If you write 10 to 15 performance reviews in a cycle, the drudgery is often the reformatting: turning scattered notes into the structure your review template demands. A performance-review assistant can take that part off your plate.

The workflow is quick to picture. Enter your role and the reviewee's role, pull up your standard questions, then dictate or paste approved, anonymized notes about specific projects, behaviors, and outcomes. The tool maps those notes into the format your process requires. You still edit for accuracy, fairness, and voice before anything enters the performance system. Once you have your source notes and review questions ready, use a prompt like this to structure the first draft:

Act as an HR performance-review drafting assistant.

Using only the approved, anonymized manager notes below, draft responses to our standard performance review questions.

Keep the language specific and evidence-based. Tie each statement to the project, behavior, or outcome documented in the notes.

Do not add information that is not provided, infer intent, assign a performance rating, or recommend a performance consequence. If the notes do not support a response, flag the gap for HR or manager review.

Standard review questions:
[PASTE QUESTIONS]

Approved manager notes:
[PASTE ANONYMIZED NOTES]

That editing step is not a nicety. Hallucinations, confident statements unsupported by the input, are a documented failure mode for language models, and their frequency varies by model, task, and benchmark. The course teaches a three-step review for exactly this: 1) check the formatting, 2) inspect the facts and completeness, then 3) correct or re-prompt rather than accepting draft one.

There is a second risk that has nothing to do with factual accuracy: employees may experience AI-assisted reviews as surveillance or impersonal decision-making. Be clear about what the tool does and does not do. The same discipline applies when a review feeds a calibration meeting or performance improvement plan, but qualified people remain responsible for the judgment, rating, and any resulting action.

The verdict: AI writes the first draft, and you write the review that gets submitted. Read every line, verify it against the underlying evidence, and make it yours before it enters the system.

3. Prepare for a pay-equity review

Pay equity is one of the highest-stakes analyses in HR, and it is one that many "AI in HR" lists ignore entirely. AI can help HR prepare for the review by identifying the questions to ask, the inputs to gather, the process steps to plan for, and common watchouts before the analysis begins.

A formal pay-equity analysis may require detailed employee-level compensation data, appropriate comparator groups, job and level information, location, tenure, performance inputs, and a defensible statistical methodology. That analysis belongs with qualified compensation, analytics, and legal professionals, not in a general generative AI workflow.

Use AI here as a preparation and planning tool. Because a general-purpose model is drawing from its training data and the context you provide, treat its output as a starting checklist rather than authoritative legal or statistical guidance. Verify requirements with your compensation, analytics, and legal partners.

Use a prompt like this to prepare for the review:

Act as an HR research assistant helping me prepare for a pay-equity review.

Create a practical checklist for HR that includes:

1. information and inputs we may need to gather
2. questions to ask compensation and analytics partners
3. questions to raise with legal counsel
4. common data-quality issues or watchouts
5. decisions that should be made before the analysis begins
6. roles and responsibilities across HR, compensation, analytics, and legal
7. a suggested process from preparation through review and follow-up

Do not analyze employee compensation, recommend individual pay changes, or make legal conclusions.

Where requirements may vary by jurisdiction or methodology, flag them for verification rather than presenting them as fact.

The verdict: use AI to help HR prepare for a pay-equity review, not to analyze employee compensation. Let qualified compensation, analytics, and legal professionals handle the underlying data, methodology, and pay decisions using approved organizational systems and tools, which may include capabilities embedded in your HR Information System (HRIS).

4. Review anonymized attrition and engagement patterns for workforce trends

Can an HR generalist use AI to explore a workforce question without a dedicated analyst? Yes, if the question is specific and the data has already been appropriately aggregated and approved for use.

Say you want to test a hypothesis: are remote employees less engaged but more productive than their on-site peers? Start with group-level metrics from your HRIS, analytics platform, or approved reporting tool, such as average engagement score, productivity measures, historical attrition rate, and work-arrangement type. Avoid uploading individual employee records simply because identifiers have been removed.

Once you have an approved, aggregated dataset, use a prompt like this to explore the pattern:

Act as an HR analytics expert and review this approved, anonymized, aggregated workforce dataset. It includes average engagement score, productivity measures, historical attrition rate, and work-arrangement type by group. No names or IDs are included.

Compare remote, hybrid, and on-site employees on these metrics and flag any group with a materially different pattern. Frame each finding as Who (the segment), What (the metrics), and Gap (the measured difference).

Use only the data provided. Do not infer causes, predict individual employee behavior, or recommend employment decisions. Flag any small sample sizes or missing information that could make a comparison unreliable.

In one course exercise, remote CAD technicians showed the highest productivity of the three groups but engagement scores nearly 40% lower than their hybrid and on-site peers. That turns a vague concern into a precise policy question. It does not explain why the difference exists or establish that work arrangement caused it.

AI enablement strategist Chiara Barden, who teaches the course's data-analysis module, calls the framing method the "insight equation": Who is the specific segment, What are the key metrics shown side by side, and what is the Gap or business impact?

"Quantifying what you're saying is always stronger than just saying a broader statement."

The ethical line matters here. AI surfaces segment-level patterns across a group. It should never output a prediction about one named person. As Barden puts it:

"AI supports HR judgment, and it analyzes patterns, not people."

Keep employee-level and potentially re-identifiable information out of this workflow. Examples include:

Best for testing a specific hypothesis with approved, anonymized, aggregated data. Treat the output as a pattern to investigate, not an explanation of cause or a prediction about an employee.

5. Personalize onboarding communications and reduce routine administrative work

Onboarding is where repetitive admin piles up fastest, and it is also where AI and automation can give back some of the most obvious hours. Instead of handing every new hire the same generic checklist, you can use AI to draft role, department, and location-specific onboarding content from approved source material, use automation to trigger and track routine tasks, and pair it with a scoped companion for policy questions.

The time savings can be substantial. IBM reports that its internal AskHR system automates more than 80 HR tasks. In 2024, it handled more than 11.5 million employee interactions and achieved a 94% containment rate for common questions. IBM also documented productivity gains in specific HR tasks between 2022 and 2024, with some improving by as much as 75%.

Even a narrow workflow adds up. In one internal course case study, connecting account provisioning across systems reduced a recurring onboarding task from about 90 minutes to roughly four minutes. The reclaimed time went back into designing the employee experience rather than clicking through forms.

To decide what is safe to automate, use the RIPE test. Score a task from 1 to 3 on four dimensions:

• Repetitive: Does it happen often in the same way?
• If-then: Can you state the rules clearly?
• Predictable: Are the inputs and outcomes consistent?
• Error-prone: Would automation reduce manual mistakes?

A score of 10 to 12 makes the task a strong automation candidate. A score of 7 to 9 is the judgment zone, where you should pilot first. Below that, keep it human.

Once you know which onboarding tasks are appropriate to personalize, use a prompt like this to draft employee-facing content:

Act as an HR onboarding assistant.

Using only the approved onboarding information below, create a personalized onboarding checklist for a new employee in [ROLE], [DEPARTMENT], and [LOCATION].

Include:
1. pre-start requirements
2. first-day priorities
3. first-week meetings and tasks
4. required systems or access
5. required training
6. key policies or resources to review
7. questions the employee should raise with HR or their manager or onboarding buddy

Do not invent policies, deadlines, benefits, system requirements, or role expectations that are not included in the source material. If information is missing, flag it under "Needs confirmation."

Approved onboarding information:
[PASTE APPROVED SOURCE MATERIAL]

Onboarding maps cleanly onto the Automate, Augment, Extend model: automate the checklist and provisioning, augment drafting and personalization where appropriate, and protect the human welcome. A 30-60-90-day plan and a warm first week are where retention starts, and no dashboard replaces them. Watch time-to-productivity as the real metric, because the sooner a new hire has access and a clear plan, the sooner they can contribute.

The verdict: use automation for rule-based setup and tracking, and use AI to draft or personalize approved onboarding content. Protect the first-week welcome, manager connection, and judgment-heavy conversations as human time.

6. Draft job descriptions from approved role information

If you try only one example on this list this week, make it this one. Drafting job descriptions is a good entry point for generative AI because the source material is usually already defined: the role purpose, responsibilities, qualifications, reporting relationship, and organizational language.

The difference between a generic and usable draft is the prompt. A weak prompt like "Write a job description for a marketing manager" gets you generic filler. A structured prompt gives the model the context, source material, format, and boundaries it needs. Use the CRAFT framework:

• Context: Give background instead of a bare command.
• Role: Assign the AI a relevant role, such as an HR job-description editor.
• Action: State exactly what you want it to produce.
• Format: Specify the output structure you want.
• Tone: Match the language to the audience, organization, and purpose.
  

Context: We are hiring a mid-level marketing manager for a 15-person team at a B2B software company.

Role: Act as an HR job-description editor.

Action: Draft a job description using only the approved role information provided below.

Format: Use Summary, Responsibilities, Required Qualifications, Preferred Qualifications, and Reporting Relationship.

Tone: Professional, clear, and direct. Avoid buzzwords such as "rockstar" and "ninja."

Do not add responsibilities, qualifications, salary information, legal requirements, or other details that are not included in the approved source material. If information is missing, flag it under "Needs confirmation."

Approved role information:
[PASTE APPROVED ROLE INFORMATION]

Before publishing, compare the draft against the approved role profile and check that the AI has not added requirements that could unnecessarily narrow the candidate pool.

The verdict: job descriptions are a lower-risk place to build HR prompting skills because AI can help structure and refine approved role information while HR retains control over the actual requirements of the job.

7. Answer routine employee questions from approved HR sources

HR teams answer many of the same questions repeatedly: where to find a form, when open enrollment begins, how a standard process works, or what an approved policy says. Those high-volume, consistent questions can be a good fit for an AI assistant grounded in your organization's approved HR documents.

The key is scope. The assistant should answer only from designated sources, identify where the information came from, and say when the available material does not support an answer. It should also have clear boundaries around questions that require interpretation, personal advice, an eligibility determination, an exception, or HR judgment.

Once you have defined the approved sources and boundaries, use instructions like these to establish the assistant's scope:

You are an HR information assistant.

Answer only general employee questions about [LIST APPROVED TOPICS] using the approved HR policies, benefits materials, and process documents provided.

For every answer:
- use only information contained in the approved sources
- identify the source document used
- say when the available documents do not answer the question

Do not give legal, tax, medical, or financial advice; interpret an employee's personal circumstances; determine eligibility; approve exceptions; or recommend what an employee should do.

If a question requires interpretation, personal information, or HR judgment, stop and direct the employee to:
[APPROVED HR ESCALATION PATH]

Good options include questions with stable, documented answers, such as where to find a policy, when a deadline occurs, which form starts a process, or what an approved plan document says. Questions about an employee's individual eligibility, leave situation, accommodation, complaint, pay, performance, or other sensitive circumstances should move to a qualified person.

Track more than how many questions the assistant resolves. Review whether answers are accurate, whether sources are current, and whether the assistant correctly recognizes when it should escalate. A high deflection rate is not useful if the tool is confidently answering questions that belong with HR.

The verdict: use AI to make approved HR information easier to access, while keeping interpretation, exceptions, personal advice, and consequential decisions with qualified people.

8. Organize interview and assessment evidence for qualified human review

What if you did not have to rewatch a two-hour interview to find the three moments relevant to the criteria your hiring team defined upfront? AI can transcribe a live or recorded interview, organize evidence against the approved rubric criteria your hiring team defined upfront, surface the relevant excerpts or timestamps, and generate a summary for a qualified reviewer.

Review this interview transcript against the following approved, job-related criteria: [criterion 1], [criterion 2], and [criterion 3].

For each criterion:
- identify the relevant evidence in the transcript
- quote or timestamp the specific response
- summarize the evidence without adding interpretation
- mark "insufficient evidence" when the transcript does not provide enough information

Do not score, rank, compare, or recommend the candidate.
Do not infer anything about the candidate that is not directly stated.
Do not use protected characteristics or other non-job-related information.

The interviewer or hiring panel still evaluates the evidence and applies the organization's approved scoring process. AI can make the review faster by showing where relevant evidence appears, but it should not determine the candidate's score or whether they advance.

There is real potential to make the process more consistent. There is also real risk if the model invents evidence, overweights irrelevant language patterns, or shapes the reviewer's judgment before they examine the underlying response. Validate the tool, obtain any required recording consent, protect the transcript, and regularly test whether it surfaces relevant evidence consistently across candidates.

The verdict: use AI to organize interview evidence and reduce review time. Keep scoring, comparison, and hiring decisions with qualified people using the organization's approved process. AI surfaces the evidence; people evaluate it.

FAQs about the use of AI in HR

The practical value of applying AI in HR depends on the boundaries around it. These answers cover the disclosure, job-impact, data, and accountability questions that should shape your first pilot.

Do I have to tell candidates or employees that AI was used in a decision about them?

It depends on the tool, location, and use case, so check the rules that apply before deployment. New York City's Local Law 144, for example, requires notice at least 10 business days before certain automated employment decision tools are used. Even where a specific notice rule does not apply, transparent communication can build trust. In a Pew Research Center survey, 66% of U.S. adults said they would not want to apply for a job if the employer used AI to help make hiring decisions.

Is AI going to replace my HR job?

The honest answer is that AI redistributes HR work rather than eliminating the need for HR judgment. It can support repetitive, high-volume tasks such as organizing information, scheduling, summarizing, and first-draft writing. Decisions about hiring, pay, performance, and discipline still need responsible people who understand the context and consequences.

Which AI use case should I try first?

Score your options instead of guessing. Rate each one on impact, feasibility, ethics, and time to value. For automation, use the RIPE test: Repetitive, If-then, Predictable, and Error-prone. A lower-risk augmentation task such as drafting job descriptions, onboarding content, or routine HR communications from approved source material usually makes a better first pilot than a use case that influences a consequential employment decision.

What employee or candidate data is appropriate to use with AI?

Do not assume data is safe to use simply because the AI tool is approved by your organization. Follow your organization's policy and rules for the specific tool, data type, and use case. Avoid entering names, employee IDs, exact compensation, medical or accommodation information, disciplinary records, identifiable free-text comments, or other sensitive employee or candidate data unless that use has been explicitly authorized for your role. Anonymized data can also be re-identifiable when groups are small or several fields are combined, so use aggregated data wherever possible.

Can AI score, rank, or recommend candidates or employees?

Higher-risk uses such as scoring, ranking, predicting, or recommending outcomes about people require much stronger scrutiny than drafting or summarizing. For most HR use cases, a safer role for generative AI is to organize evidence against predefined criteria and leave the evaluation, scoring, and consequential decision to qualified people using the organization's approved process.

Who is responsible if an AI-assisted decision is wrong or biased?

Legal responsibility varies by jurisdiction, contract, and use case. Operationally, your organization should act as though it retains ownership of the decision. Document the process, validate the tool, monitor outcomes, preserve meaningful human review, and involve legal counsel in consequential uses. A vendor recommendation is not a substitute for your organization's judgment or obligations.