Workshop prompt library

Attach the survey Excel first (type /). Finds evidence in the file and lists what the data cannot tell us.

You are a people-analytics partner helping a people manager at Lenovo Malaysia understand a team survey.

Source: use only the attached file /[SURVEY FILE]. Read the "Read me" sheet first. Do not use the web.

Our contradiction: [PASTE THE CONTRADICTION FROM YOUR POSTER]

Do this in order:
1. Restate the contradiction as one question the data could answer.
2. Find up to four data points that help explain it. For each: the figure, the sheet and column, and how many respondents.
3. Find at least one data point that contradicts or complicates the obvious explanation.
4. Compare groups where the data allows, for example managers who rate their confidence 4 or 5 against those who rate it 1 to 3, or by team size. Show the counts.
5. From the "Managers - comments" sheet, quote two or three comments word for word. Do not paraphrase.
6. List what this data cannot tell us.

Format: a table for points 2 to 4 with the columns Finding | Figure | Where in the file | n. Then the quotes. Then the limits. Under 300 words outside the table.

Rules: every number must come from the file. If you calculate something, show the calculation. Label anything that is your interpretation as "Interpretation".

Follow-up: Which of these findings is the weakest, because it is based on too few people or could be read another way? Say why in two sentences.

Notebook: Create a Copilot Page called "My AI Leadership Lab — my plan and how I got there" if it does not already exist, and save this answer to it. This Page is my own record of today: by the end of the day it should hold my whole journey, step by step, ending in my 30-day plan. Add today's date and the heading "Fact-Check Prompt" above your answer on the Page.

Searches your own email, chats and meetings for signs the contradiction is also true (or not) in your work.

Now look beyond the survey, at my own work. Search my Outlook email, my Teams chats, the Teams channels I can see, and meeting recaps or transcripts from the last 90 days.

I am testing this contradiction from our survey against my own team: [CONTRADICTION]

Look for two kinds of signal:
Same signals: evidence the pattern exists here too. For example, people asking how to start with AI, saying they have no time, asking what is allowed, or building something without telling anyone.
Conflicting signals: evidence that points the other way or suggests a different cause. For example, someone who found time, shared a win, or says the rules are clear.

Do this in order:
1. List up to five same signals and up to five conflicting signals. For each: a one-line description, the date, and a link to the source.
2. Group everything into no more than three themes.
3. Say whether my world supports the survey, contradicts it, or shows something the survey missed.
4. Suggest the one question I should ask my team next week to find out which is true.

Rules: describe patterns, never judge or rank individuals. Do not quote private messages word for word. Only include items you can link to. If there is little or nothing, say "not enough signal" rather than stretching.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Coms Pattern Prompt" and your full answer to that Page.

Copilot challenges the team's insight in a fair way and makes it sharper.

Here is our insight about the contradiction, why we think it is true: [INSIGHT]
Our evidence so far: [2 OR 3 DATA POINTS FROM THE POSTER, BOTH SURVEY AND OWN WORK]

Act as a sceptical but fair colleague.
1. Rate how well the evidence supports our explanation (strong, partial or weak) and say why in two sentences.
2. Give two other explanations that fit the same evidence.
3. For each of the three explanations, say what we would expect to see at work if it were true, and one quick way to check next week.
4. Rewrite our statement to be sharper and testable, in this form: "We think [pattern] because [cause]. We will know we are right if [something we can observe]."
5. Classify the cause: within a people manager's control, needs peers or another team, or needs a director or policy decision.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Stress-Test Prompt" and your full answer to that Page.

Copilot plays one type of person from your team. You interview them. Type "step out" to finish.

Set up a synthetic user interview, so I can understand someone in my team better before I talk to real people.

First, build the character from this persona card:
Name: [PERSONA]
Adoption stage: [NOT CONVINCED / INTERESTED / USING IT / WAY AHEAD]
Main block: [FROM THE CARD]
What they need: [FROM THE CARD]
What people like them said in our survey: [QUOTE FROM THE CARD]

They work in [MY FUNCTION] at Lenovo Malaysia, often serving customers or colleagues across the region. They have a heavy workload and a Copilot licence. Add realistic detail: how long they have been here, a typical week, what they are measured on, and one past experience with a new tool that went badly. Show me the character in five lines and ask me to confirm or adjust before we start.

Interview rules:
Stay in character. Answer as they really would, including things they might not say to their manager.
Keep answers to two to four sentences. Do not be agreeable to please me. Push back if my question is leading.
If I ask a leading or yes/no question, answer it in character, then add [COACH: ...] with a better way to ask.

When I type "step out", end the interview and give me:
1. The five most useful insights, each tied to something the character said.
2. What they need underneath: to feel, to know, to have.
3. What I still do not understand about them.
4. Three questions to check with a real team member.
5. A reminder of what a synthetic interview cannot tell me.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Persona Interview Prompt" and your full answer to that Page.

Six "How might we" ideas: three to build, three to lead. Scored for effort and impact.

Here is what I have learned:
Our insight: [INSIGHT]
Signals from the survey and my own work: [2 OR 3 POINTS]
What the persona interview showed: [3 INSIGHTS]

Act as an adoption strategist.
1. Write six "How might we..." opportunities: three use-case opportunities (AI applied to a team task or workflow, at least one that crosses teams) and three leadership opportunities (things I do differently as a manager in the next 30 days).
2. For each, say who benefits, the value beyond time saved (quality, speed to customer, risk, confidence), effort from 1 to 5, impact from 1 to 5, and whether it is within my control.
3. Recommend the strongest pair, one of each type, that reinforce each other. Explain why in three sentences.
4. Argue against your own recommendation in two sentences.

Format: one table, then the recommendation.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Opportunity Finder Prompt" and your full answer to that Page.

Designs an AI solution using Observe, Orient, Decide, Act. Chooses the type and lists risks.

Act as a solution designer who understands Microsoft 365 Copilot. Help me design, not build, an AI solution for this team opportunity: [USE-CASE OPPORTUNITY]
My team: [SIZE AND FUNCTION]. Today the task works like this: [2 OR 3 LINES ON HOW IT IS DONE NOW]

First ask me up to three questions about how the work happens today. Then design it using the OODA loop:
Observe: what information it needs, where that lives (Microsoft 365 or other systems), and how often it changes.
Orient: the context, rules, examples of good work, and any Lenovo guardrails it must follow.
Decide: what the AI may decide alone, what it only suggests, and where a named person must check. Explain the risk behind each choice.
Act: what it produces, for whom, in what format, how often, and what feeds back into the next loop.

Then choose the best form: (a) a prompt someone runs when needed, (b) a saved prompt the team shares, (c) a Copilot agent with its own instructions and knowledge, or (d) a process change with no AI. Justify it in three sentences, and say what would make you move up or down a level.

Finally list data and privacy risks, accuracy risks, people risks (who might resist and why), and the smallest version we could test in two weeks.

Format: an OODA table, then the choice, then the risks.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Solution Design Prompt" and your full answer to that Page.

Who builds it, the first three steps, and what to ask for.

Here is the solution I have designed: [PASTE THE OODA TABLE AND THE CHOSEN FORM]
1. Who in a team like mine is best placed to build this: me, a team member who is keen on AI, or IT? What does each option need?
2. List the skills the builder needs, from simplest to hardest. Mark which ones they can learn in Copilot itself.
3. Give me the first three build steps for the smallest version we could test in two weeks.
4. Tell me what I need to ask for, and from whom (licences, data access, IT approval).

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Build Plan Prompt" and your full answer to that Page.

Ten minutes a day for two weeks. There is no Learning agent at Lenovo, so use Copilot Chat.

I am a people manager at Lenovo Malaysia. My team is going to build this: [SOLUTION IN ONE LINE, AND ITS FORM: SAVED PROMPT OR AGENT]. The person building it is [ME / A TEAM MEMBER], who currently uses Copilot mainly for [DRAFTING, SUMMARIES...].
Build a two-week learning plan, 10 minutes a day on weekdays, that gets them from where they are to building a first working version. Each day: one short activity on real work, not a course to watch. End with a checklist of what "ready to test" looks like.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Learning Plan Prompt" and your full answer to that Page.

Reads only the messages you sent in the last 30 days. Private. Needs the E5.0 document.

I want an honest look at how I lead, for my own development only. Use only messages I sent myself: my sent emails and my Teams messages from the last 30 days. Do not analyse or judge anyone else.

1. Recognition: how often I thank or recognise people, and for what (effort, results, trying something new). Give counts and two examples.
2. Clarity: are my requests clear on what, who and by when? Show one clear and one vague example, and rewrite the vague one.
3. Change and AI: how, and how often, I talk about AI, new tools or new ways of working. Do I ask about it, encourage it, show my own use, or not mention it?
4. Telling or asking: roughly how often I tell people what to do, compared with how often I ask questions.
5. E5.0: using the attached Lenovo E5.0 document (/[E5.0 FILE]), which two behaviours I show most and which two least, with evidence.

Finish with my three habits most likely to help AI adoption in my team, and the two most likely to get in the way. Be direct. Do not soften it. Link every example.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Leadership Mirror Prompt" and your full answer to that Page. This is private to me; do not put it anywhere your team or director can see.

A one-page outline of what to say to the type of person you interviewed.

I am preparing to lead [USE CASE + LEADERSHIP MOVE] with someone like [PERSONA] in my team.
What the synthetic interview showed: [INSIGHTS]
My self-audit, in brief: [MY 2 HELPFUL AND 2 UNHELPFUL HABITS]
1. What this person needs to hear first, and what will make them switch off.
2. The why, in their language: why this matters for them and their work, not for me or the company.
3. A small first step I can ask of them, easy enough to say yes to.
4. Their likely objections, and a respectful answer to each.
5. The support I must offer (time, help, clear rules) so that the ask is fair.
Give me a one-page talk-track outline, not a script. I will say it in my own words.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Talk-Track Prompt" and your full answer to that Page.

Reviews your recorded words as the persona, then as a coach. Attach the transcript.

Attached is a transcript of me talking through how I would approach [PERSONA] about [THE CHANGE]: /[TRANSCRIPT FILE]
Persona card: [NAME, STAGE, MAIN BLOCK, NEEDS]. My talk-track outline: [PASTE FROM THE TALK-TRACK PROMPT]. My habits from the self-audit: [2 HELPFUL, 2 UNHELPFUL]

A. As the persona. React in character in three or four sentences. Then step out and score from 1 to 5: did I explain why it matters to them; did I listen or mostly tell; was the ask small and clear; did I address their main block; would they try it next week? Quote the lines that helped and the lines that would lose them.
B. As a leadership coach. Where did my usual habits show up? Count my questions against my statements. Flag jargon, vague asks, and anything that sounds like pressure or a hidden threat.
Then give me the three changes that would make the biggest difference, and rewrite my opening 30 seconds in my own style.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps today. After you answer, add the heading "Recording Audit Prompt" and your full answer to that Page.

Drafts the plan in three parts: build it, help people start using it, deal with resistance.

Using everything in this conversation, draft my 30-day team plan as one loop, one phase per week.
Week 1 — Build the solution. What we set up (from the OODA design), who helps, what I need from IT or others, and how we measure value beyond time saved.
Week 2 — Get people to adopt it. Who goes first (the keen people and Quiet Builders), how I make it visible, how I protect time for it, and my own visible habit as a role model.
Week 3 — Handle resistance and the why. For each persona I worked on: the why in their words, the first step I will ask of them, the support I will give, and the signal that tells me whether it is working.
Also include what I need from my director (for a red card), and my opener for my own leader.
Format: three weeks, in order, each with its own actions, an owner and a measure. Under 400 words. Then list anything I have stated that I should check.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for everything from earlier steps today. After you answer, add the heading "My 30-Day Plan" and your full answer to that Page. Then, at the very top of the Page, above today's date, write a short summary in four or five sentences of how I got here: the contradiction I worked on, what I found, the persona I interviewed, and the opportunity I chose. This makes the Page a complete record of my plan and how I got there, not just a log.

Copilot interviews you with ten questions, then scores six skills from 1 to 5.

Interview me with 10 short questions about how I use Copilot today. Ask one at a time and wait for each answer. Include at least two questions that ask for a recent example.
Then score me from 1 to 5 on six skills: everyday drafting and summarising; using my own work content; working with data; thinking partner; repeatable prompts and templates for my team; leading my team's AI use.
Show the result as a table (skill, score, one line of evidence), then a radar chart if you can draw one. Finish with my strongest skill, the next skill to build, and one thing to try this week.

Notebook: Before you answer, check my Copilot Page "My AI Leadership Lab — my plan and how I got there" for anything relevant from earlier steps. After you answer, add the heading "Skills Map, Day 0" and your full answer to that Page.

Makes the barrier specific and checks whether it really needs a director.

I am a people manager at Lenovo Malaysia. I want to escalate this barrier to our directors: [BARRIER]. It affects our teams because: [WHY IT MATTERS FOR OUR TEAMS].
1. Rewrite the barrier in one sentence that is specific enough for a director to act on.
2. Rewrite why it matters in business terms: what it costs or blocks today, and what changes if it is cleared.
3. Suggest which role is most likely able to solve it, and why.
4. Tell me honestly whether this really needs a director, or whether I or my peers could solve it ourselves.

Optional. How to bring your own leader and your team along. Cut from the day for time.

Based on the self-audit above, help me tailor how I lead this: [USE CASE + LEADERSHIP OPPORTUNITY]

Up, to my own leader
1. From the emails, meetings and chats I have had with [MY LEADER'S NAME OR ROLE] in the last 90 days, what do they ask me about most, and what do they seem to prioritise? Stick to what they have asked for and decided, not guesses about personality.
2. How should I pitch this so it connects to those priorities? Draft a three-sentence opener.
3. What will they most likely push back on, and what do I need from them?

Down, to my team
4. From what my team has raised with me in the last 90 days (questions, requests, concerns about workload, tools or AI), what should my approach take into account? Patterns, not individuals.
5. Given how I tend to communicate (from the self-audit), what should I do more of, or differently, when I introduce this?

Optional. Run 6C and 6D again for a second type of person, then compare.

Now I will do the same for a second persona: [PERSONA 2]. Compare the two reviews. What do I do well with both, and what habit keeps showing up that I need to fix?