Juliette's Learning Tribe

AI skills for project managers: the tasks to change first

Learn AI as a project manager on the paperwork you already do: meeting notes, status reports, RAID logs and plans. What to hand over, and what to own.

  • ai-and-work
  • future-fluency
  • project-management
An Idea to Take With You
One next step, with you.

The fastest way to learn AI as a project manager is on the paperwork you already do every week. Start with meeting notes and status reports. Then move on to the RAID log and the plan. Keep the judgement calls for yourself.

You don’t need a course on machine learning. You need one task, done with AI, checked properly, every week for a month.

Most project managers have started

The Association for Project Management asked 1,000 UK project professionals about this, with Censuswide. Its findings, published in March 2026, say 27% have AI fully embedded in their project work.

The uses near the top of the list are the everyday ones. Admin like summarising documents. Reporting and dashboards. Risk forecasting.

When they asked about the skills project managers will need most, ethical decision-making and professional judgement came joint top. I love that. The tool does more of the typing, and the person matters more, not less.

The five tasks to change first

Pick one. Do it with AI for four weeks before you add another.

  1. Meeting notes and actions. Paste in your rough notes. Ask for the decisions, the actions, an owner and a date for each. Then check every owner and every date against what was actually said. This is where it invents things.
  2. The weekly status report. Give it last week’s report and this week’s changes. Ask for a draft in the same format. You set the RAG status yourself. Never let it pick the colour.
  3. The RAID log. Your risks, assumptions, issues and dependencies. Ask it to find duplicates, items with no owner and actions past their date. Then ask what risks a project like yours usually carries that aren’t on the list. It is good at spotting gaps. It knows nothing about your organisation.
  4. The plan. Ask it to break a deliverable into tasks and list the dependencies. Then ask it to argue with your estimates. It doesn’t know how fast your team works, so treat its numbers as questions.
  5. Stakeholder updates. Same facts, three versions: one for the sponsor, one for the team, one for a supplier. Read each one as that person would. Fix the tone before it goes.

Illustration of three colleagues at a table. One woman points to a simple three-step flow diagram in an open notebook while the other two listen.

What you must still own

AI can draft almost everything on a project. It can’t be accountable for any of it.

  • The bad news. If a milestone is slipping, you say so, in your words, to the right person first.
  • The RAG status. Amber or red is a judgement about people, money and politics. That’s your call.
  • The trade-offs. Scope, time or cost. Somebody has to choose, and explain why.
  • The relationships. The sponsor who is unhappy needs a conversation, not a better-written email.
  • Your name on the report. If it’s wrong, it’s your report that was wrong.

The weakness: it sounds right when it’s wrong

AI sounds exactly the same when it is wrong as when it is right. A status report that is late is annoying. A status report that is confidently wrong is much worse, because people act on it.

So the checking is the skill. Read every draft as if a bright new starter wrote it. Look for the date nobody agreed, the owner who wasn’t in the room, the risk that has quietly turned into an issue.

In August we found the levels on our job descriptions were wrong. AI had written them, and they read perfectly. But every entry-level role in the 2,371 we checked came out at Level 3 to 4. A Junior Accounts Assistant, with no one reporting to them, sat at Level 3 to 4. It should have been Level 2. The same page said entry level. Two days later we had re-levelled 59,198 roles. I like a big mistake. You can see it, so you can fix it.

Keep project data safe

Projects are full of things that shouldn’t leave the building. Supplier prices, staff names, client plans.

The UK’s National Cyber Security Centre says not to put sensitive or confidential information into public AI tools. Its guidance is here. Use the AI tool your organisation has approved, and follow its rules. There’s more on this in is it safe to use ChatGPT at work.

How to learn AI as a project manager

  1. Choose the task you dread most each week. If you can’t choose, start with the status report.
  2. Write down how long it takes now. You’ll want the number later.
  3. Do it with AI for four weeks. Check every line. Note what kind of wrong it was.
  4. Save the instructions that work. A saved prompt or a template is your first small automation.
  5. Measure again. Is it faster? Is it better? What did you do with the time you got back?

The professional bodies are moving too. APM launched a three-part module on prompting for project professionals alongside that research. There’s a longer version of this method in how to learn AI skills for work.

Start from your own role

Zavmo is a personal AI tutor built around your work. You start with your role, and the lessons use it.

Have a look at a real role page, like Project Manager, Business Transformation and Change. It shows what the job involves and the skills it needs, and it’s free to read. There are many more under project management.

Then try the first module of Future Fluency: AI skills for work. It’s free, with no card. Bring this week’s status report.