LO
Project manager — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Status reports, minutes and summaries
Automating✓ Evidence-backedWriting the weekly status report, the meeting minutes, the executive summary and the notes that keep everyone working from the same picture.
Summarising what was said and what changed is what language models do best, and meeting and tracking tools now draft these documents from the data they already hold. In PMI's 2024 survey of project professionals who already use generative AI, 53% used it to summarise and review content.
The surveys record what people say they use AI for or gained from it, often among those who already use it; they do not measure whether organisations have stopped needing a person to write these documents.
Building and updating the plan
Being augmented✓ Evidence-backedBreaking the work down, sequencing it, estimating it and re-planning when something slips.
Tools can draft a work breakdown, propose a schedule and flag the dependencies a change breaks. Knowing which estimate is optimistic, which team is overloaded and which date the sponsor will not move is context the project manager holds.
The surveys record benefits people report from task and schedule automation; they do not measure how plans are now built or who builds them.
Tracking risks, costs and changes
Being augmented✓ Evidence-backedKeeping the risk log, watching the budget, and assessing what each requested change will do to scope, time and money.
Dashboards and models can spot a cost overrun or a slipping milestone early. Deciding which risk deserves attention, and saying no to a change the sponsor wants, is a judgement someone has to own.
The UK's major-projects authority uses its tool across a portfolio, not inside a project's own risk log; it does not show how risk and change decisions on a project are now made.
Aligning stakeholders and negotiating trade-offs
Still human-led≈ Platform inferenceGetting sponsors, teams and suppliers to agree on scope, dates and priorities — and to keep agreeing when things change.
The work is persuading people with different interests to accept a trade-off, often without authority over them. A tool can prepare the options; accepting the cost of one of them is a conversation between people.
This rests on how the work is done, not on a measurement of how often these negotiations now involve AI.
Leading the team and clearing blockers
Still human-led≈ Platform inferenceKeeping people moving: noticing who is stuck, removing what blocks them, and holding the team together when the project gets hard.
Tools can show that a task is late; finding out why, and changing what a person or another team does about it, depends on trust and on being accountable for the result.
This rests on how the work is done, not on a measurement of how team leadership is changing.
Deciding where AI output enters the project
New task≈ Platform inferenceChecking AI-drafted plans and reports before they go out, and setting what the team may hand to AI tools and what it may not.
When status reports, plans and summaries are drafted by tools, someone has to be answerable for what they say, and someone has to decide which project data may go into which tool. On most projects that falls to the project manager.
No source yet measures how much of a project manager's time now goes to checking AI output or setting rules for it; this is an inference from the direction of the work.