AI in HR Presentation: An Outline With Examples and Checks

An AI in HR presentation works best when it leads with the business problem, not with the technology. Slides that open with a list of AI capabilities tend to prompt skeptical questions about whether any of those capabilities actually apply. Slides that open with "we currently spend 15 hours a week sorting applications for roles with more than 80 applicants" give an audience something concrete to engage with.

The outline below is a starting point you can adapt. Use an AI assistant to draft each section from your own notes, then check every factual claim against a primary source before you present. Tools like fenbs let you keep the approved briefing prompt and the source links in one shared place, so updates to the slide deck start from the same reviewed materials rather than re-researching from scratch.

What this outline covers

• Slides 1-2: The business problem your HR team faces today
• Slides 3-5: Specific use cases with evidence and sources
• Slides 6-7: Risks, review steps, and approval requirements
• Slides 8-9: A proposed pilot and how you will measure it
• Slide 10: Questions and next steps
• A draft prompt for each section
• A pre-presentation checklist

Slides 1-2: The business problem

Slide 1 states the problem in measurable terms. How many hours does the task take? How often does output need correction? What does the delay cost in recruiter time or candidate experience? One or two concrete numbers are more persuasive than a general claim about inefficiency.

Slide 2 explains why this problem matters now. That might be a change in hiring volume, a new compliance requirement, or a team reduction that left the same workload with fewer people. Avoid the argument that AI matters because it is new; that is not a business reason.

I need to write two slides for an HR presentation. Slide 1 should state our problem with [task] in measurable terms: [current volume, time cost, error rate]. Slide 2 should explain why this matters now: [reason]. Write each slide as a heading and three bullet points. Use plain language. Do not include any statistics I have not provided.

Slides 3-5: Specific use cases with evidence

Each use case slide covers one application: what the AI does, what a person must still do, and what the evidence shows. Three use cases is enough for a first proposal. More than five use cases in a single presentation tends to dilute rather than strengthen the argument.

Every factual claim about what a tool can do needs a primary source. If you cite a vendor case study, note that it comes from the vendor. If you cite published research, give the journal and year. Audience members who know the field will check, and an unchecked claim that turns out to be wrong damages the whole presentation. For what the research actually shows, AI in HR case studies summarizes documented evidence with primary source guidance.

Write one slide about using AI for [specific HR task]. Structure it as: what the AI does, what a person must review or approve, and what a published source says about the outcome. I will add the source myself. Keep each point to one sentence. Do not invent statistics.

Slides 6-7: Risks, review steps, and approvals

A presentation that does not cover risks reads as incomplete to any legal, compliance, or senior HR reviewer. Slide 6 names the risks specific to your proposed use case: disparate impact if screening is involved, employee data privacy, vendor accountability, and what happens if the AI produces a wrong output.

• Disparate impact: AI tools trained on historical data can replicate historical bias. The EEOC's technical assistance on automated employment decision tools covers this at eeoc.gov.
• Employee data privacy: Note whether employee or candidate data leaves your organization and under what terms.
• Review process: Name who reviews AI output before it reaches an employee or candidate and how errors are corrected.
• Transparency: Several jurisdictions require that applicants know AI is being used. Check current requirements for your locations.

Slide 7 covers the approval chain: who can approve the pilot, who must be informed, and what sign-off looks like. Including this slide shows the audience that you have thought past the technology to the organizational process.

Slides 8-9: Proposed pilot and success measures

Slide 8 describes the pilot in concrete terms: one task, one team, a defined time period, and the baseline you will measure against. A pilot scoped to one role type for six weeks is easier to approve than a company-wide deployment proposal.

Slide 9 lists the success measures you agreed on before starting. What result would justify continuing? What result would cause you to stop? Writing these in advance, and including them in the presentation, is important: it shows that the evaluation is not after-the-fact and that you are prepared to stop if the evidence does not support continuing.

Slide 10: Questions and next steps

End with the three decisions you need from the room: approval to proceed, a named sponsor, and a review date. Leave space for questions. The questions an audience asks will often reveal assumptions you need to address before the pilot starts.

Pre-presentation checklist

• Every factual claim about a tool's performance cites a primary source, not a roundup or a summary.
• Risk slides cover disparate impact, data privacy, review process, and transparency requirements.
• The pilot has a baseline, a time period, and pre-written success criteria.
• The presentation does not promise outcomes the tool has not demonstrated in your context.
• Someone who was not involved in building it has read it and identified the weakest claim.

How to do this with fenbs

Use fenbs to keep the presentation materials, prompts, and source links in one shared project.

1. Create a free account and add a project for the presentation.
2. Add a task for each slide group: the business problem, use cases, risks, and pilot proposal.
3. Store the draft prompts in Skills and prompts so any connected AI assistant can use them. Link each prompt to the primary sources it depends on.
4. Record presentation decisions in Decisions and rules: which claims are approved, which sources are verified, and who reviewed the risk slide.
5. Before the presentation, check Links and listings to confirm every source URL is still live.