From the PMI talk · Partnering with AI
PM AI Prompt Library
How to use: Approved tools only. Replace bracketed placeholders. Cut what’s wrong out loud before you send.
Meeting & Communication
Summarize meeting notes
Summarize these meeting notes for a project manager who was not in the room.
Include only what is supported by the notes:
• Decisions made
• Open questions
• Risks or concerns raised
• Next steps (only if clearly stated)
Rules:
• Quote or paraphrase tightly — do not invent
• If something is unclear or missing, write “Unknown — not in notes”
• Flag assumptions separately from facts
• Plain language. No filler.
Meeting notes → action table
From these meeting notes, create an action table with columns: Action | Owner | Due | Status | Notes.
Rules:
• Only include actions that were clearly assigned or agreed
• If owner or due date is missing, put “TBD — confirm”
• List every ambiguity under Ambiguities at the top
• Do not invent owners, dates, or commitments
• Status = Open unless the notes say otherwise
Draft stakeholder email
Draft an email to [stakeholder / group] about [topic].
Context: [1–3 bullets]
Goal: [inform / request decision / ask for help / reset expectations]
Voice: plain first-person. No jargon. No blame. One clear ask.
Give me:
1. Subject line
2. A short body
3. One softer version and one more direct version
Do not invent facts, dates, or commitments I did not provide.
Cut / edit my draft
Here is my draft:
[paste draft]
Edit it like a skeptical PM whose name will be on it.
• Cut what is wrong, fluffy, overconfident, or unsupported
• For each cut, tell me why in one short line
• Tighten language — spoken and clear, not corporate
• Do not add new claims I did not make
• Return: (1) edited draft (2) list of cuts with reasons
Project & Portfolio
Review a status update
Review the attached project update as a skeptical PM.
Identify:
• Risks
• Decisions required
• Missing information
• Questions leadership is likely to ask
Then draft a concise executive summary I can stand behind.
Rules:
• Cut fluff and unsupported claims
• Separate facts in the materials from your inferences
• If something is unclear, say so — do not paper over it
Analyze risks
Review the attached project materials. Identify risks I should be watching.
For each risk:
• What it is
• Evidence from the materials (quote or cite)
• Why it matters now
• What would make it worse
• A practical next check to discuss with the team
Label each item Known (in materials) or Speculative (inferred).
Do not invent severity scores unless the materials support them.
Identify dependencies
From these notes, decks, and tickets, list dependencies that could delay delivery.
For each:
• Upstream / downstream
• Who owns it (or “Unknown”)
• What happens if it slips
• Confirmed vs inferred
Call out hidden dependencies. Do not invent owners or dates.
Surface conflicting information
Compare these sources: [list files / paste excerpts].
Surface conflicts only — dates, owners, status, scope, or commitments that do not match.
For each conflict:
• What disagrees
• Where each version came from
• What I should verify with a human before publishing status
If sources agree, say so briefly. Do not invent a third “truth.”
Consolidate risks, issues & dependencies
From these PowerPoint / status documents, create one consolidated table of Risks, Issues, and Dependencies.
Columns: Type | Item | Source deck | Notes
Rules:
• Note the source for every row
• Do not merge items that are clearly different
• Flag likely duplicates (same item, different wording)
• Do not invent severity, owners, or dates not in the materials
Leadership
Draft written executive update
Draft a concise written executive status update from the attached notes and materials.
Audience: [senior leadership / VMO / steering committee]
Length: half a page or less
Tone: plain, confident, no fluff
Structure:
1. Where we are
2. What changed since last update
3. Risks / blockers that need attention
4. Decisions needed (if any)
Do not overclaim progress. If something is unclear, say so.
Do not invent metrics, dates, or commitments.
Spoken briefing bullets
From these materials, draft spoken briefing bullets for [audience].
Rules:
• No more than 5 bullets total
• Each bullet is one breath — write for speaking, not slides
• No filler adjectives, no technical deep-dives
• Only include a next step if it is real and supported
• Flag anything I must verify before I say it out loud
Talking points
Create talking points for a [15 / 30]-minute conversation with [audience] about [topic].
Include:
• Opening frame (1–2 sentences)
• 3–5 points I should hit
• Likely questions and an honest draft answer for each
• What I should not overpromise
Write for speaking. Plain first-person. Do not invent facts I did not provide.
Decision briefing
Draft a one-page decision briefing.
Decision needed: [state it]
Options: [A / B / C or unknown]
Context: [paste or attach]
Structure:
• Decision required
• Why now
• Options and tradeoffs
• Recommendation — label clearly as DRAFT FOR MY EDIT
• Risks of delaying
I will own the recommendation. Do not present speculation as fact.
If options are incomplete, say what is missing.
Change impact assessment
We are considering this change: [describe].
Assess impact on:
• Scope / timeline / dependencies
• Stakeholders who need to know
• Communications required
• Risks if we proceed vs. if we wait
Call out what you cannot assess from the materials.
Do not invent impacts, dates, or stakeholder reactions.
Story across sources
Read these files. I need to brief leadership.
What is the through-line? Propose a short narrative outline:
• What we committed to
• Where we are
• What is in the way
• What I need from them (if anything)
Outline only — no slides.
Flag anything I should verify before I say it out loud.
Do not invent a cleaner story than the sources support.
One prompt worth memorizing
Review the attached project update. Identify:
• Risks
• Decisions required
• Missing information
• Questions leadership is likely to ask
Then draft a concise executive summary I can stand behind.
Cut fluff. Do not invent what is not in the materials.
Analyze → draft → you edit. That pattern is the craft.
Accelerate the work. Don’t dilute the craft.
