A co-pilot, not a replacement
Most project managers do not lose their week to big decisions. They lose it to writing: status reports, meeting summaries, risk logs, and emails that say the same thing three ways for three audiences. That is the work ChatGPT and similar tools do well.
What these tools cannot do is lead. They do not know your sponsor's priorities, the history between two departments, or which vendor always delivers late. They cannot be accountable for a decision. The best way to think about them is as a quick, well-read assistant who has never met your stakeholders: useful for a first draft, never the final word.
The working rule: draft, check, decide. Let AI produce the first draft. Check every fact, name and number against your own sources. Then make the decision yourself, in your own words.
How to use the prompts
Each use below comes with a prompt you can copy. Replace the parts in [square brackets] with your own details. The more context you give (the project, the audience, the format you want), the better the draft. Remove names, contract values and anything confidential before you paste.
Initiating
Start projects on a clear footing
1. Turn a vague request into a draft charter
Projects often begin as an email or a hallway conversation. Paste the request and ask for a draft charter. The value is not the document itself but the gaps it exposes: missing objectives, no success measures, no named sponsor. Those gaps become your questions for the kickoff.
Here is a project request: [paste request]. Draft a one-page project charter with purpose, objectives, high-level scope, out of scope, success measures, key stakeholders, assumptions and constraints. Then list the questions I must answer before the charter can be approved.
2. Map stakeholders and what they care about
Describe the groups affected by the project and ask what each one is likely to worry about. You will get a starting stakeholder register and a sense of where resistance may come from. Then correct it with what you actually know about the people involved.
For a project that will [describe the change], list the likely stakeholder groups. For each one, give their main interests, likely concerns, level of influence and a suggested way to engage them. Present it as a table.
Planning
Plan faster without skipping the thinking
3. Break the scope into a first-cut WBS
A work breakdown structure is tedious to start from nothing. Ask for a draft with deliverables and work packages, then refine it with your team. The team's corrections are where the real estimates come from. You can build the final version in the free PM Artifacts Lab, which draws the WBS tree for you.
Create a work breakdown structure for [project and main deliverables], three levels deep, with WBS codes. Mark the work packages that usually carry the most estimating uncertainty and explain why.
4. Stress-test the schedule
Instead of asking AI to build your schedule, ask it to attack one. Share the main milestones and durations and ask what could make them wrong. It is a quick way to find optimistic assumptions before your sponsor does. For a numeric view of schedule risk, try the Monte Carlo schedule simulator.
Here are my milestones and durations: [paste list]. Act as a sceptical reviewer. Which estimates look optimistic, which dependencies are missing, and what three questions should I ask the team before I commit to these dates?
5. Widen your risk list
Everyone has blind spots. AI is good at listing the risks that are common to a type of project: vendor dependence, data migration, key-person loss, low adoption. It cannot know the risks that live inside your organization. Use it to widen the list, then score each risk and give it an owner in your register.
List 15 risks for a [type of project] in [industry], grouped by category. For each, give the cause, the possible effect, an early warning sign and one response. Leave probability and impact blank so my team can score them.
Executing
Communicate more, write less
6. Write one update for several audiences
Executives want decisions and status at a glance. The team wants priorities and blockers. Clients want progress and next steps. Write the facts once, then ask for a version for each audience. You keep control of the facts; AI only changes the framing.
Here are this week's facts: [paste bullet points]. Write three short updates: one for executives (status, decisions needed, top risk), one for the team (priorities and blockers) and one for the client (progress and next steps). Do not add any facts that are not in my notes.
7. Turn meeting notes into decisions and actions
Rough notes become a clean summary in seconds. Read it before you send it: AI sometimes gives an action to the wrong person or turns a discussion into a decision that was never made.
Turn these meeting notes into a summary with three sections: decisions made, action items (owner and due date) and open questions. If an owner or date is unclear, write "to confirm" instead of guessing. Notes: [paste notes].
8. Rehearse a difficult conversation
Before telling a sponsor that the date will slip, practise. Ask the AI to play a sceptical sponsor and push back on your message. It helps you find the weak points in your argument and prepare clear options.
Play the role of a demanding project sponsor. I need to tell you that [the issue] and recommend [your option]. Push back the way a busy executive would. After five exchanges, tell me where my message was weak and how to make it stronger.
Monitoring and controlling
Make sense of the numbers and the changes
9. Explain variances in plain language
Share anonymized figures and ask for a plain-language explanation of what they mean and what to look into next. Always recalculate the numbers yourself, because AI can misread a column or invent a total. The free earned value calculator gives you CPI, SPI and forecasts you can trust.
My project has these figures: planned value [PV], earned value [EV], actual cost [AC], budget at completion [BAC]. Explain in plain language what they say about cost and schedule, what the likely final cost is, and which three questions I should ask the team. Show your calculations so I can check them.
10. Assess a change request
When a change request arrives, AI can help you think through its full impact before it reaches the change board. It will not know your real capacity, but it will remind you of the knock-on effects people forget: testing, training, documentation and other teams.
A stakeholder has asked to [describe the change] on a project that is [current status]. List the possible impacts on scope, schedule, cost, quality, risk and other teams, and the questions the change board should ask before deciding.
Closing and growing
Capture the learning, and keep learning
11. Draft lessons learned and the closure report
Lessons learned are often skipped because the team has already moved on. Collect quick notes from the retrospective and ask AI to organize them. Make sure each lesson ends with something specific the next project should do differently.
Organize these retrospective notes into lessons learned: [paste notes]. Group them into what went well, what did not and what to change. Turn each lesson into one specific, actionable recommendation for future projects.
12. Learn a concept and test yourself
AI is a patient tutor. Ask it to explain a framework with an example from your own industry, then ask it to quiz you. If you are preparing for the PMP exam, confirm key points against the PMBOK® Guide and practise with the free PMPORT exam simulator.
Explain [concept, for example earned value or Kanban] using an example from [your industry]. Then ask me five questions, one at a time, and tell me after each answer whether I am right and why.
Know the limits
What AI gets wrong
| Weakness | What it looks like | What to do |
|---|---|---|
| Invented facts | A confident date, figure, name or reference that does not exist. | Check every fact against your own sources before you use it. |
| Generic answers | Risk lists and plans that would fit any project. | Give more context and ask it to be specific to your situation. |
| No organizational context | Advice that ignores politics, history and culture. | Treat the output as a starting point and apply your judgment. |
| Out-of-date knowledge | Old versions of standards, tools or exam content. | Confirm against the latest official source. |
| Agreeing too easily | It supports whatever you suggest. | Ask it to argue the other side or to find flaws in your plan. |
Governance
Five rules for safe use
Getting started
A 30-day starter plan
Week 1: Status reports
Write this week's update from your bullet points with use 6. Note how long it took compared with usual.
Week 2: Meetings
Summarize two meetings with use 7. Check every action owner before sending.
Week 3: Risks
Run use 5 before your next risk review and bring the new ideas to the team to score.
Week 4: Review
Add up the time saved and the errors you caught. Keep what worked, drop what did not, and agree simple team rules for AI use.
The bottom line
AI will not make anyone a better project manager on its own. It gives back time, and what matters is how you spend it: more time with stakeholders, more thought on the decisions that matter, and less on formatting documents. Draft with AI, check with care, and lead in your own voice.