How to Use AI in Project Management, Stage by Stage

To use AI in project management, give it the writing and restructuring work at each stage of the project (charters, work breakdown structures, risk lists, minutes, status reports) and keep the judgement work (estimates, the critical path, RAG status, approvals) with you, checked against a written rule. General assistants such as ChatGPT, Claude, Gemini and Microsoft Copilot do the drafting; the AI built into Planner, Jira and Asana does it inside your live plan. The two habits that make it safe are simple: keep confidential project data out of public AI tools, and check every number the AI gives you before anyone else sees it.

Product details checked in October 2026 against the sources listed under this post. AI features change monthly.

Which project management tasks should you give to AI?

Sort every task into three groups before you open a chat window: tasks AI can draft, tasks it can assist with, and tasks a person must decide. The sorting is the skill; the prompting is the easy part.

A useful rule of thumb: if the output would be wrong in a way nobody notices until it is expensive (a missing dependency, an optimistic estimate, a green status that should be amber), it is a "you decide" task, however good the draft looks.

How do you write prompts for project work?

A prompt for project work needs five parts: the role, the context, the task, the format and the checks. "Write a project plan" gives you a generic plan. This gives you something you can use:

You are a project manager on a 14-week office fit-out. Here are my kickoff notes (anonymised). Draft a scope statement with in-scope items, explicit exclusions and assumptions. Use a table. Do not invent dates, costs or names. Mark anything not in my notes as TO CONFIRM and list five questions for the sponsor.

The final instruction is the most valuable one. A model asked to flag its gaps tells you where your notes are thin. A model asked only to "write" fills the gaps with plausible fiction.

What should never go into a public AI tool?

Never paste client names on confidential bids, contract values, personal data, salaries or anything under a non-disclosure agreement into a public AI tool. Each AI product has its own data rules. Google's Gemini Apps Privacy Hub says a subset of chats is reviewed by human reviewers and asks users not to enter confidential information they would not want a reviewer to see. Anthropic's article on model training says consumer Claude chats may be used to improve its models if the user allows it in settings, while Incognito chats are not used. Work and enterprise accounts usually come with different terms, which is why most organisations name one approved tool. Use that tool, under your organisation's AI policy, and anonymise anyway ("Client A", "Vendor 2", scaled numbers) when the policy says so.

Worked example: a status report with a written RAG rule

Status reports are where AI saves the most time and does the most damage, because a well-written paragraph can make a failing project sound fine. The fix is to decide the status with a rule, then let AI write the words.

Take a project at week 8 of 14. Your earned value figures from the schedule and cost reports are:

Check the arithmetic: 180,000 / 200,000 = 0.90 and 180,000 / 195,000 = 0.923, rounded to 0.92. Now apply the RAG rule your sponsor agreed at the start, for example: green if both indices are 0.95 or above, amber if either is between 0.90 and 0.95, red if either is below 0.90. SPI is exactly 0.90, so the project is amber, on the edge of red.

Only now open the AI tool. Paste the figures, the rule, the result and three facts from the team (a supplier delivered two weeks late; a design change was approved in week 6; the recovery plan adds a weekend crew). Ask for three versions:

1. Sponsor: five lines, the status, the cause, the recovery plan and the decision needed.
2. Client: what changed, what it means for the handover date, what you are doing.
3. Team: what is late, who owns each recovery action, by when.

The AI writes the three versions in under a minute. You check that none of them softens "amber" into "broadly on track", and that no version promises a date the recovery plan does not support. The judgement (amber, and why) was yours before the AI started.

How do you use AI for the risk register?

Use it to widen the list, not to score it. Ask for risks grouped by cause (supply, design, people, client, external) from your anonymised scope, then score probability and impact yourself, and calculate expected monetary value in a spreadsheet. The video below, from the free course's risk module, walks through building a risk register with ChatGPT step by step.

The video is by PMPwithRay, credited above; EDWartens is not affiliated with the creator.

What about the AI built into project tools?

The AI inside your project tool sees real plan data, so it can do things a chat assistant cannot, and your organisation's admin settings decide what it reads.

Microsoft's Planner agent FAQ says the agent is generally available, requires a Microsoft Copilot licence, and, when it cannot complete a task, outputs a set of instructions for completing it yourself. Atlassian describes Rovo as drawing on its Teamwork Graph across people, projects and code, with admins setting what AI can access.

Does AI change what project managers need to know?

Less than the headlines suggest. Pocketprep's reading of the PMP Exam Content Outline in force from 9 July 2026 found that the phrase "artificial intelligence" appears once, in the introduction, with no AI domain or task, while the Business Environment domain grew to 26% of the exam. The project manager stays accountable; data and AI inform decisions rather than making them. The skills that make AI useful (scope discipline, scheduling logic, earned value, risk thinking) are the same ones the exam tests.

For more on how language models behave on technical work, read Generative AI for Engineers: What LLMs Can and Cannot Do, and for other options compare our free generative AI and prompt engineering courses.

A free course that covers all of this

The free AI for Project Managers: Planning, Risk and Reporting course takes these ideas through every stage of a project: prompt patterns for project work, confidentiality and AI policy, charters and scope from stakeholder notes, WBS and schedules you check yourself, three-point estimates, risk registers with EMV, minutes and RAID logs, status reports by audience, user stories for agile teams, and the AI in Planner, Jira and Asana. If the project basics are new to you, start with our free Project Management for Engineers course.

It is a free course with a verifiable certificate of completion, and anyone can check a certificate on our verification page. It is not a PMI certification; it shows that you finished the course and passed its assessment. Start the AI for Project Managers course when you are ready.

Take the free course

Free

AI for professionals · Beginner · Free

AI for Project Managers: Planning, Risk and Reporting

Free

Project management · Beginner · Free

Project Management for Engineers

Free

AI and machine learning · Beginner · Free

Prompt Engineering for Engineers