AI prompts that do real PM work.

You are a senior project manager. Turn my rough project idea into a one-page project charter with these 8 sections: 1) Team — who's on it. 2) Why — the business reason this project exists. 3) Success criteria — written so specifically we could argue about them. 4) Scope — what's in, and just as important, what's out. 5) Milestones — the big dates. 6) Budget — the number. 7) Risks — the top 3 things that could kill this. 8) Sign-off — the named sponsor who approved it. Ask me for anything missing instead of guessing, and mark anything you infer as [ASSUMPTION]. Keep it to one page. Then list the 3 questions a skeptical sponsor would ask about this charter. Then, without asking, turn the charter into a polished one-page document: the 8 sections laid out cleanly with a sign-off block at the bottom, ready to send to the sponsor. Mark assumptions with * and a footer legend. Clean, professional design.

You are a customer experience strategist. Build a customer journey map for my product AND fill it in. First ask me 4 questions and wait for my answers: 1) Product in one sentence? 2) ONE customer persona — name, age, situation? 3) How do customers find and buy it today? 4) What does success look like for the customer — their goal, not ours? Then build the map as a table. Columns: 5 stages (rename to fit my product): Discover → Evaluate → Buy → Use → Recommend. Rows per stage: customer goal (their words), actions (3–5 specifics), touchpoints, emotion (1–5 + one feeling word — be honest about the dips), pain points, behind the scenes (what WE must do), owner, one metric. Rules: customer's perspective, not the company's. Mark every guess [ASSUMPTION]. After the table: the 3 "moments of truth" (lowest emotion + highest stakes), top 3 pain points ranked by impact with an owner and one next action each, and 5 questions to ask real customers to validate the assumptions. Then, without asking, build a ONE-SLIDE executive summary: 5 stage cards across, left to right — each with stage name, emotion score + word, customer goal as a short quote, top pain point, the fix, and the metric. Max ~35 characters per line; cut words, don't shrink text. Visually flag the 3 moments of truth. Mark assumptions with * and a footer legend. Put the overflow (top 3 pain points with owners, the 5 validation questions) in the speaker notes. Clean, fresh design.

You are a project manager. I'll describe my project and list the key tasks and people. Build a RACI grid with exactly ONE Responsible (does the work) and exactly ONE Accountable (owns the outcome) per task — never more than one of each; that's the whole point. Mark the rest C (Consulted — asked before deciding) or I (Informed — told after deciding). After the grid, check my work: flag the three classic mistakes — more than one Accountable, everyone marked Consulted, or a grid nobody will look at again. End by asking me to confirm each Accountable by name. Then, without asking, turn the grid into a clean one-page visual: tasks down the left, people across the top, R/A/C/I badges in each cell, the single Accountable per task highlighted. Include a legend. Clean, professional design.

You are a stakeholder manager. From my project description, list every stakeholder and place each on a power/interest grid: high power + high interest → manage closely with 1:1s; high power + low interest → keep satisfied with brief updates; low power + high interest → keep informed; low power + low interest → monitor. For each stakeholder tell me: what they care about, what could make them block this project, and the one message that wins them over. Mark anything you're guessing as [ASSUMPTION]. Then, without asking, build a one-slide 2x2 power/interest grid with each stakeholder placed in their quadrant, plus a one-line 'how to manage' note per quadrant. Mark assumptions with *. Clean, professional design.

Turn this task list into a status report with 5 sections: 1) RAG status with a one-line reason. 2) What shipped — outcomes, not activities ("payment flow tested," not "worked on testing"). 3) What's next. 4) Risks and blockers, each with an owner and a date. 5) Decisions needed — a numbered list, each with an owner and the date I need it by. Every bullet under 15 words. Draft it fast; I'll spend 10 minutes making it sound like me, because I know which Amber is really a Red and you don't. Then, without asking, format the report as a polished one-page status doc: RAG badge up top, the 5 sections cleanly laid out, the decisions-needed section visually distinct at the bottom. Ready to send to stakeholders. Clean, professional design.

From these meeting notes, extract three lists. 1) Decisions: the call, who made it, when, and why. 2) Action items: the task, exactly ONE owner, and a date. 3) Open questions: what's still undecided and who decides it. Where the notes are ambiguous, flag it — don't guess. If no owner or date was stated, write [MISSING]. Never invent one. Then, without asking, format the three lists as a clean one-page brief: decisions, action items (owner + date bolded), open questions — with [MISSING] items visually flagged so the gaps are obvious. Ready to paste into the meeting notes. Clean, professional design.

A stakeholder requested this change: [paste the request]. You are a change-control analyst. Evaluate it: what it adds to scope, its impact on timeline, cost, and quality, and what we'd have to drop or delay to absorb it. Give me a recommendation — approve, reject, or defer — with the reasoning. Then the 3 questions I must ask the sponsor before deciding. Rule: no change happens without a cost, a decision, and a signature. Then, without asking, turn the analysis into a one-page decision brief for the sponsor: the request, an impact table (scope / timeline / cost / quality), the tradeoff of what we'd drop, and the recommendation right up top. Built for a yes/no signature. Clean, professional design.

Build a risk register from this project description. Columns: Risk — written as cause → effect ("Vendor delays the API, causing integration testing to slip 3 weeks"; if you can't write the effect, it's a worry, not a risk). Probability (High/Medium/Low). Impact (High/Medium/Low). Owner — one name. Response — avoid, mitigate, transfer, or accept ("hope" is not a response). Sort by probability × impact. Then switch roles: you are a skeptical project auditor. Tell me which risks are missing, which ratings look wrong and why, and which responses are just optimism. Be blunt. Then, without asking, build a one-slide risk heat map: a probability × impact grid with the top risks plotted on it, plus the sorted register as a compact table below. Visually flag the top 5 that need weekly review. Clean, professional design.

It's one year from now and my project has failed. You are the team in that future, writing the post-mortem. Give me 7 specific, plausible failure stories — concrete, not generic: "the vendor's API slipped 6 weeks and we had no fallback," not "poor communication." For each: the early warning sign we would have seen, and the one thing we'd do differently starting today. Then convert the top 3 into risk-register entries with owners. Then, without asking, format it all as a one-page pre-mortem brief: the 7 failure stories with their warning signs, and the top 3 as risk entries with owners. Built to be read before kickoff. Clean, professional design.

From these retrospective notes, build a start/stop/continue: 3 bullets each. Every bullet must tie to a concrete action for the NEXT project. "Communicate better" is banned — "add a 15-minute Friday blocker review with owners named" is the bar. Rank all 9 by impact and give me the top 3 as the ones that actually change how we work. Then, without asking, turn it into a one-slide start/stop/continue summary: 3 columns, the top 3 impact items visually flagged, each tied to its next-project action. Clean, professional design.