AI in project management: project charter, risks, status report
Written with AI and checked against the sources by a second AI. As of November 16, 2026. More in the notes on AI.
Projects rarely fail for lack of templates. They fail because of unclear goals, overlooked risks and reports that paint everything green. An AI helps you think of the right questions and finish texts faster. Estimating, deciding and reporting honestly are still up to you, together with your team.

Key points
- For the project charter, first have the AI ask you questions, especially about what is out of scope.
- The AI suggests work packages and risks as a checklist. Your team estimates duration and effort.
- In the status report, you set the traffic light color yourself. The AI only does the wording.
- Retrospectives evaluate processes, not individual people.
- Documents from client projects, quotes and contracts stay out of private AI accounts.
Where an AI can help in a project
| Task | What the AI contributes | What you and your team contribute |
|---|---|---|
| Project charter | Questions, outline, wording | Goals, boundaries, budget, decision |
| Work packages | A suggested structure | What really needs doing, duration, who is responsible |
| Risk list | Checklist of typical risks | Assessment and countermeasures |
| Status report | Text from your bullet points | Traffic light color, figures, decisions |
| Updates for stakeholders | Versions for different groups | Facts and approval |
| Retrospective | Sorting the notes | What you want to change |
In the trial lesson you put your first instruction together and get feedback. No account, in a few minutes. Start the trial lesson
The project charter: questions first, then text
Rhea is an IT project manager at a municipal utility. She is to introduce a new ticket system for customer service. The assignment from management fits into three sentences. Before she turns it into a project charter, she has the AI interview her.
I need to write a project charter. The project in three sentences: [text].
First ask me at most ten questions I need to answer for a good project charter.
Ask explicitly about non-goals, meaning what is not part of the project, about dependencies on other projects, and about who signs off at the end.
Write the charter only after I have answered, and use only my answers.The question about non-goals makes Rhea think: is the connection to the billing system part of it or not? She settles this with management before the project starts, and so avoids an argument in the third month.
Let the AI suggest work packages, estimate effort yourself
Pavel is production manager at a maker of plastic packaging. A production line is to move to another hall. He has the AI suggest a work breakdown structure, that is, a breakdown of the project into subtasks and work packages.
Suggest a work breakdown structure for moving a production line to another hall on the plant site.
What we know: [machines as abbreviations, dates, constraints].
Divide it into at most six subtasks with three to eight work packages each.
Don’t estimate any duration or effort.
Mark work packages that are often forgotten with [easily forgotten].The marked packages include things like the inspection of the new power connections and informing customers about a possible pause in deliveries. Afterwards, the people who do the work estimate the duration of each package. A figure from the chat would be an average of other people’s projects, not an estimate for Pavel’s hall.
Risks: looking back from the end
A risk list gets better if you don’t ask “What could go wrong?” but pretend the project has already failed and look for the reasons. This exercise is called a pre-mortem. An AI is not part of the team, so it can name reasons nobody on the team likes to say out loud.
Imagine our project [short description] has failed one year after it started.
Name twelve possible reasons, sorted by technology, people, money, time and suppliers.
For each reason, write an early warning sign that would let us recognize it.
Don’t rate anything. We will rate likelihood and impact as a team.Pavel brings the list to the next team meeting. The team crosses some items off, adds two points only insiders know about and rates the rest in a simple matrix of likelihood and impact. For the most important risks, it decides who watches which warning sign.
Status report: you set the traffic light
Keira is a business analyst and reports to the steering committee every two weeks. She writes bullet points and has the AI turn them into a short report. One rule matters to her: the AI must not gloss over anything.
Turn my bullet points into a status report with these sections: overall status, done, next steps, risks and issues, decisions I need.
Overall status: [red/amber/green], take it over as it is, don’t change it.
Describe problems as clearly as they are in the bullet points. Don’t soften anything.
One page at most. Dates and figures only from my bullet points.
Bullet points:
[bullet points]AI texts often sound more confident than the situation is: “Test postponed, cause unclear” quickly becomes “Test will be carried out shortly.” So Keira reads the sections on risks and decisions with particular care and compares them sentence by sentence with her bullet points.
Meetings that end with decisions
Many project meetings end with “We’ll take another look at that.” So Rhea prepares her steering committee meetings in a way that every agenda item ends with a question that can be answered with yes or no. The AI helps her phrase the items that way.
Here are the topics for our next steering committee meeting, 60 minutes.
Turn them into an agenda. For each item: goal (information or decision), time in minutes, who presents (role), and for decisions the exact question to be decided.
If a topic is too unclear for a decision, write [clarify beforehand: …].
Plan five minutes at the end for open items.
Topics:
[bullet points]Rhea sends the agenda round two days in advance, with the documents for each decision. Whatever was marked [clarify beforehand], she settles in a preliminary talk. For the minutes afterwards, there is a separate guide: Meeting minutes with AI.
One update, three versions
Idris is an event manager and is planning a machine builder’s presence at a trade fair. The same news, the booth will be smaller than planned, has to go to management, to the booth staff and to the booth builder. Each group needs something different.
News: [facts].
Write three short messages:
1. Management: what changes, what it costs or saves, which decision is needed. At most 80 words.
2. Booth staff: what changes for their shifts and tasks. At most 80 words.
3. Booth builder: which dimensions and dates change, with a request for a new quote. At most 100 words.
All three messages contain the same facts. Don’t make up any new ones.Idris puts the three versions side by side and checks that the same dimensions and dates appear everywhere. With several versions, discrepancies creep in easily.
Retrospective: evaluate processes, not people
At the end of the project, the team collects what went well and what didn’t. An AI can group many sticky notes into topics. For that, copy the notes without names. A sentence like “Flynn slept through the test” becomes “Test was started too late.”
Here are the notes from our project retrospective, without names.
Group them into at most six topics. For each topic: what went well, what didn’t, and one question we should clarify in advance in the next project.
Don’t rate any people or roles.
[notes]Evaluating the performance of individual employees is a different matter. If there is a works council (Betriebsrat, the elected employee representation), a company may as a rule only introduce software that can monitor behavior or performance with its consent. A retrospective on processes doesn’t need that, as long as it sticks to processes.
What stays out of the chat
Projects are full of documents that don’t belong in a private AI chat: quotes and prices from suppliers, contracts, customer data, plans for new products. A confidentiality agreement often forbids this explicitly. And the law only protects a trade secret if the company itself takes reasonable care to keep it secret.
If the AI is already built into the project tool, it usually sees everything in the project. Whether you use it, and for which projects, is for your company to decide.
Check before you pass it on
- Compare goals, non-goals and dates with the person who commissions the project.
- Every figure comes from your documents or from an estimate by the team.
- Risks and issues are as clear as in your notes.
- Put several versions of the same message side by side and compare the facts.
- No names in retrospectives, no confidential documents in the chat.
For your next project
You learn the most if you deliberately do one step differently in your next project, for example by starting the project charter with questions from the AI. This cycle of instruction, result and check is what the learning app “KI können” trains, each time with your own AI. The trial lesson is the quickest way to try it once yourself.
Sources
- German Works Constitution Act (Betriebsverfassungsgesetz), Section 87 Rights of co-determination (in German) (opens in a new tab)
- German Trade Secrets Act (Gesetz zum Schutz von Geschäftsgeheimnissen, GeschGehG), Section 2 Definitions (in German) (opens in a new tab)
Information as of November 16, 2026. Laws and offers can change; this article is not legal advice.
Read more
Try it now
The trial lesson with examples for IT project manager. No account, a few minutes.
Start the trial lesson