AI in Project Management: Less Status Reporting, More Judgment

AI is starting with the coordination work

Much of a project manager’s week goes to collecting status, updating schedules, writing up meetings and chasing follow-ups. That is the work AI handles best today. McKinsey’s October 5 research found that coordinator roles, project management among them, grew 1.5 to 2 times faster than line roles over five years. That growth is now the target.

The tools are changing quickly. UC Today reported on October 7 that OpenAI’s new always-on agents can own a project, with rules that set which actions run on their own, which need approval and which are blocked. Work management vendors such as Asana are repositioning around the same idea.

Faster reports are not faster projects

In McKinsey’s 2026 survey, 80% of respondents reported individual productivity gains from AI, but only 37% reported a positive effect on EBIT. A status report written in ten minutes instead of two hours does not deliver a project any sooner. Measure what the PMO exists to improve: how long decisions wait, how accurate forecasts are and how often milestones slip.

Write the rules before the agents run

Decide which project actions an AI tool may take on its own, which need a person to approve and which are off limits. Updating a task status is low risk. Moving a deadline, reallocating budget or messaging a client is not.

Most large companies are still early here. KPMG’s September AI Pulse found that 49% have defined high-risk uses where an agent may not decide alone, and 74% include cost reviews when approving AI work. Wharton’s October 13 report found only 17% of departments let an agent finish a task without human sign-off at each step. Every agent should have a named person who owns what it does.

The project manager’s job shifts to judgment

McKinsey describes a product team that moved from pods of eight to ten people to hybrid teams of four to six supported by agents. Smaller teams with more automated coordination put more weight on the parts of the job AI does poorly: weighing tradeoffs, reading stakeholders, deciding when to escalate and owning the result.

Some executives expect bigger changes. One told Caixin on October 8 that most middle managers will disappear as AI takes over routine coordination. Whether or not that happens, project managers who can show judgment and accountability will be harder to replace than those whose value is the weekly report.

Build what you can run

Gartner predicted on September 29 that 70% of enterprises will abandon agentic AI built for them by vendor engineering teams by 2028. Its advice applies to any PMO automation: name an executive sponsor accountable for the outcome, require knowledge transfer and an exit plan, and set the balance between automation and human oversight before launch.