How Am I Supposed to "Just Use AI for That?" (Live Workshop)
When leadership says "just use AI for that" with no guidance, HR ends up guessing, over-trusting drafts, or avoiding AI altogether.
What We Covered
- Intern Mental Model: Treat AI like a smart intern. Give context, iterate fast, stay skeptical, capture system instructions when something works.
- Persona SuperPrompt: Role + goal → ask what persona to assume → let it interview you → turn the result into reusable instructions.
- Planning before producing: Structure the ask (constraints, audience, output format) before you want polished output. Lindsay's frame: anticipate bias, legal, and data risks in the plan, not after send.
- McKinsey dual mandate (chat): HR must transform its own work while leading org redesign. Credibility comes from going first.
Live demos — prompt patterns
Each live demo used the same contrast: a rushed "just use AI" paste, then a constrained "actually use AI" version.
Demo | Rushed ask | Built ask (key constraints) | Watch for |
|---|---|---|---|
| A — Email | "Read this thread and tell me how to respond." | Use only the thread · flag unstated · 3 decisions, 5 risks, 3 reply options | Leaders may have already communicated dates the summary misses |
| B — Manager email | "Write an email about using AI for HR work." | Interim guardrails · under 150 words · no tool/data promises · disclaimer | Vague mandates produce over-promising drafts |
| C — Meeting notes | "Turn these into action items." | Table: Action Item, Owner, Deadline · TBD when missing | Models invent owners or skip deadlines |
Bonus pattern (not run live): Policy FAQ — attach your doc, prefix with Use only this document as source., then ask an edge case the doc does not answer. Without a source, the model guesses.
Demo A — email thread
Use only the thread below. Don't invent facts. Flag anything unstated as needs verification.
Read this email thread and give me: 3 decisions that need to be made, 5 key risks, and 3 options for how to reply.
[paste email thread here]Demo B — manager email
Write a friendly and informative email for all managers about interim expectations for using AI on employee-facing drafts while our official AI use policy is still in Legal review. Highlight three practical guardrails managers can use this week. Under 150 words. Supportive tone. No promises about which tools are approved, what data is allowed, or that AI output can be sent without human review. Add a disclaimer: check the manager hub for updates when Legal publishes the final policy.Demo C — meeting notes
Read these meeting notes. Create a table: Action Item, Owner, Deadline. If owner or deadline isn't stated, mark TBD.
[paste meeting notes here]Northwind dummy scenarios from the live session ship in the workshop follow-up email.
Attendee coaching moment
Eran Arbel asked about summarizing a large batch of senior-hire resumes. Jonathan walked through the Persona SuperPrompt to build a TA analyst persona; Lindsay stressed plan-first: define fields, bias risks, and verification before batching.
Member Voices
Jonathan Flower
"Talk to it just like an intern."
Lindsay Hansen
"Don't invent any facts."
Laurie Gartrell
"We want to make sure it's credible, before we just hand it over."
Session Materials
- HR's dual mandate in the AI era (McKinsey)
- Imagine Talking to an Intern Guide (PDF)
- Personal Meeting Notes Prompt Builder Guide (PDF)
Persona SuperPrompt
I am a [your role] and I am working on [goal or problem]. What persona should you assume to help me? Assume this persona and start asking me questions.50 prompts starter
Give me 50 practical AI prompts that will help me work faster, smarter, and more strategically.Go deeper — October 2026 cohort
Today's session was a sample of how we teach HR teams to use AI with guardrails, not hype. The AI for HR Mastermind premium cohort starting in October goes deeper: live coaching, applied challenges, custom assistants, and a community of HR practitioners building real workflows.
Join the October 2026 cohort →
Try this week
- Run one demo prompt above on real (non-sensitive) work.
- Try the Persona SuperPrompt at the start of a work block.
Date
Jul
09
Thursday, July 9, 2026
4:00 PM - 5:00 PM UTCRegistration is unavailable.