Facilitator Guide · Workshop Syllabus

AI for Admin Teams

A practical, platform-agnostic program for everyday work — 1 kickoff + 4 weekly sessions, delivered remotely.

60 min live 60 min async ~20 participants Remote
Click any section header to collapse or expand it.
i Program at a glance5 sessions · the weekly rhythm ›
0
Kickoff
Demystify & first win
1
Prompting
Prompts that work
2
Documents
Email, docs & notes
3
Data
Spreadsheets & organizing
4
Responsible use
Privacy + capstone
The weekly rhythm: 60 minutes live + 60 minutes async. Each live session is a tight teach-and-taste; the real practice happens in a structured one-hour mission applied to participants' own work.
1 How the program worksThe 60 / 60 model ›
60
MINUTES LIVE
Teach, demo, one breakout, share-out
60
MINUTES ASYNC
A structured mission on real work

Post-work primary, pre-work minimal

What makes AI stick is applying it to a real task you already own — so every session ends with a hands-on mission. The only pre-work is a 10-minute account setup.

The loop, every week

  1. Learn — 60-min live session (teach + demo + one breakout + share-out).
  2. Apply — 60-min async mission on the participant's own (sanitized) tasks.
  3. Share — the next session opens with quick mission share-outs.
2 Remote delivery logisticsRun of the room ›
Roster & breakouts20 participants = 4 breakout rooms of 5. Every breakout has a concrete deliverable and a named reporter.
Staffing1 facilitator + 1 producer (manages chat, breakouts, timing). A producer is essential for remote delivery.
Keep it engagingPolls, chat "waterfall", reactions, and a shared Prompt Cookbook the cohort builds across all 5 weeks.
Safety ground ruleNo real confidential data in tools during the workshop — use sanitized or dummy examples only.
CamerasOn for kickoff & breakouts; optional during demos.
MaterialsOne shared "Prompt Cookbook" doc the cohort adds to every session.
Tool stance: pick one free, widely-available chat assistant for live demos so everyone sees the same screen — but tell participants explicitly that the skills transfer to any tool. Keep the teaching platform-agnostic.
3 Before we beginPre-work + program objectives ›

Pre-work — before the kickoff only (~10 min)

  • Create a free account on the chosen tool and confirm you can log in.
  • Send one throwaway prompt: "Explain what you can help me with in my admin job."
  • Answer a 1-question survey: the repetitive task you'd most like to spend less time on.

Program learning objectives

1Explain what AI tools can and can't reliably do — and spot hallucinations.
2Write effective prompts using instruction, role and few-shot techniques.
3Apply AI to core admin work: email, documents, notes, scheduling and data.
4Evaluate and verify AI output before trusting or sending it.
5Use AI responsibly — protecting sensitive information and following policy.
0 Session 0 — Demystify & First WinKickoff · 60 min live ›

Learning objectives

  • Describe in plain terms what AI tools do — and their limits.
  • Understand the "no confidential data" ground rule.
  • Produce one genuinely useful output in the first hour.

Run of show

0:00Welcome + how the program works
0:10What AI is & isn't (great vs. wrong-answer demo)
0:22Ground rules: privacy & verify everything
0:30First Win exercise
0:45Chat share-outs + 2 volunteers
0:55Assign Mission #1, preview Week 1

In-session exercise

"Rewrite my email" — turn a blunt sample email into something warm and professional. Alt: "Two truths & a hallucination" — ask a factual question and spot what it got wrong.

Mission #1 · ~60 min

Get comfortable: ask 5 questions about your job, do one low-stakes real task end-to-end, then redo it asking AI to "improve its answer."

1 Session 1 — Prompting That Actually WorksWeek 1 · 60 min live ›

Learning objectives

  • Apply the three core techniques — instruction, role and few-shot prompting.
  • Iterate a weak prompt into a strong one.
  • Save reusable prompts to the shared Cookbook.

The three prompting techniques

These are the heart of the session — teach each one with a live demo, then a quick admin example.

TechniqueWhat it isAdmin example
1. Instruction Give a clear, specific instruction — state the task, the format, the length and the audience. "Write a 3-sentence reminder email to staff about Friday's report deadline — friendly but firm."
2. Role Tell the AI who to be so it adopts the right expertise and tone. "Act as an experienced executive assistant. Draft an agenda for a 30-minute budget review meeting."
3. Few-shot Show 1–3 examples of what "good" looks like so it matches your style and format. "Here are two meeting invites I've written: […]. Now write one for next week's onboarding session in the same style."
Then: the iteration loop. Show how one prompt improves across three passes — vague → specific → constrained — so participants see that the first answer is a starting point, not the finish line.

Run of show

0:00Mission #1 share-outs (chat waterfall)
0:05Technique 1 — Instruction prompting: live demo + admin example
0:13Technique 2 — Role prompting: live demo + admin example
0:20Technique 3 — Few-shot prompting: live demo + admin example
0:27The iteration loop: vague → specific → constrained
0:33Breakout: Prompt makeover (~13 min)
0:46Reporters share best prompt → Cookbook
0:55Assign Mission #2

In-session exercise

Prompt makeover (breakout): improve a deliberately vague prompt using all three techniques, then compare outputs. Alt: "Role roulette" — same task, three different "act as…" roles.

Mission #2 · ~60 min

Build your prompt toolkit: write & test 3 reusable prompts for weekly tasks, refine the weakest through 2 rounds, post your best to the Cookbook.

2 Session 2 — Documents, Email & Meeting NotesWeek 2 · 60 min live ›

Learning objectives

  • Draft and refine documents and emails end-to-end.
  • Summarize long content and shift tone by audience.
  • Turn messy notes into clean minutes + action items.

Run of show

0:00Mission #2 share-outs
0:05Demo: messy notes → structured minutes + action items
0:15Demo: summarize long content into a 5-bullet brief
0:23Tone-shifting: same message for peer / exec / client
0:33Breakout: Notes to Minutes + follow-up email (~12 min)
0:45Share-outs + "what would you double-check?"
0:55Assign Mission #3

In-session exercise

Notes → Minutes (breakout): turn a fake meeting transcript into minutes, action items and a follow-up email. Alt: "Tone ladder" — one message for three different audiences.

Mission #3 · ~60 min

Real documents: convert real (sanitized) notes into minutes, draft a document you owe someone, then verify every name, date and number.

3 Session 3 — Data, Spreadsheets & Getting OrganizedWeek 3 · 60 min live ›

Learning objectives

  • Extract and structure information into tables and lists.
  • Get help writing and explaining spreadsheet formulas.
  • Draft trackers, checklists and schedules from plain English.

Run of show

0:00Mission #3 share-outs
0:05Demo: unstructured text → clean table
0:13Demo: "explain this formula" + "write a formula that…"
0:22Demo: build a tracker/checklist from plain English
0:30Breakout: Mess to Table (~13 min)
0:43Share-outs + "where could this go wrong?"
0:53Assign Mission #4 + capstone briefing

In-session exercise

Mess → Table (breakout): turn a blob of unformatted info into a clean structured table. Alt: "Formula helper", or plan a 12-person offsite from a one-line brief.

Mission #4 · ~60 min

Data + start your capstone: do one real data/organizing task, then begin a genuine capstone task end-to-end, saving a quick before/after.

4 Session 4 — Responsible Use + CapstoneWeek 4 · 60 min live ›

Learning objectives

  • Apply privacy, confidentiality and verification habits.
  • Recognize one simple automation opportunity.
  • Present a real AI-assisted task to peers.

Run of show

0:00Responsible use: red-flag scenarios
0:15What's next / light automation tour
0:25Capstone shares in breakouts (rooms of 5)
0:40Room nominees present to the cohort
0:50Cookbook recap, 30-day commitment, close

In-session exercise

Red-flag scenarios (poll/breakout): is it OK to paste this here? What's the safer move? Alt: "Spot-the-error" — find the subtle mistakes in an AI output.

Program close

Pick one task to keep doing with AI for the next 30 days, and revisit the cohort Prompt Cookbook whenever you start something new.

⏱ The async hour~60 min between sessions · the four missions ›

Each mission is a structured hour — not a "try it once." Steps are timed so participants know it's a real, bounded commitment.

Mission #1 — Get comfortable

15'Ask the tool 5 questions about your job; note one great answer and one wrong/vague one.
25'Pick one low-stakes real task and do it with AI start to finish.
15'Redo the same task asking the tool to improve its first attempt.
5'Jot one sentence: was it faster? What surprised you?

Mission #2 — Build your prompt toolkit

15'List your 5 most repetitive weekly tasks.
30'Write and test a reusable prompt for 3 of them (instruction / role / few-shot).
10'Refine the weakest result through 2 rounds of iteration.
5'Post your single best prompt to the shared Cookbook.

Mission #3 — Real documents

20'Take one real (sanitized) set of notes → produce clean minutes + action items.
20'Draft a real document you owe someone; refine tone for the actual audience.
15'Verify: fact-check every name, date and number against the source.
5'Note one thing AI got wrong that you caught.

Mission #4 — Data + start your capstone

20'Use AI on one real data/organizing task (clean a list, get a formula, draft a tracker).
20'Begin your capstone: a genuine admin task completed with AI end-to-end.
15'Finish it and save a quick before/after to show.
5'Write your 1-line capstone summary for Session 4.
★ Five facilitator principlesKeep these front of mind ›
  • Practical-first — every session ends with something usable Monday morning.
  • Real work over theory — missions use participants' own (sanitized) tasks.
  • Verify always — make "trust but check" a habit, not a footnote.
  • Protect data — no confidential information in tools during the workshop.
  • Build together — grow one shared Prompt Cookbook across all five weeks.