In 2019, Gartner predicted that AI would eliminate 80% of traditional project management tasks by 2030. That prediction is still being quoted today, in fresh 2024 and 2025 industry reports, as though it were new. Six years into a ten-year forecast, and almost nobody has checked whether it’s actually holding up.
That is the real story of AI in project management. Not a revolution arriving on schedule. A five-year-old headline still doing the work, because nobody checked it.
What Was Supposed to Happen, and Didn’t
The last wave of “project management will never be the same” content leaned on blockchain, VR, and IoT as the technologies set to redefine the discipline within five years. None of them did. Blockchain-based initiatives that made real institutional bets, Maersk and IBM’s TradeLens, the Marco Polo trade finance network, the Australian Securities Exchange’s blockchain settlement system, all shut down or were abandoned, mostly for the same reason: a very slow, expensive database dressed up as a breakthrough. VR fared no better in mainstream project work. The hardware cost stayed high, the meetings stayed awkward, and the “shared virtual workspace” pitch never found a use case that beat a decent video call and a shared document.
None of that means the underlying instinct, that project management tools were about to change meaningfully, was wrong. It means the specific technologies picked to carry that change were wrong, which is a different and more useful lesson: hype cycles pick the wrong vehicle more often than they pick the wrong destination.
What Actually Changed Instead
AI adoption inside project management tools has genuinely accelerated, and the numbers are real rather than aspirational. The Project Management Institute’s own research found the share of practitioners using generative AI for more than half their project tasks, what the report calls “Trailblazers,” nearly doubled in a single year, from 20% to 37%. The most common uses are unglamorous and exactly where you’d expect an assistant to add value: cleaning and organising project data, spotting trends across documentation, summarising reports that used to eat an afternoon.
Specific tools have followed the same pattern. Microsoft’s Planner now ships a “Project Manager Agent” that reads Teams meeting transcripts, extracts the decisions actually made, and builds a workback schedule from them, useful, specific, and reported independently rather than only through vendor marketing. The people actually building and studying these tools are consistent about what they are and are not. Dr. Te Wu of PMO Advisory puts the honest ceiling on it: “AI will get you to 85, 90 and 95% accuracy for cost estimates, risk and schedules right out of the box.” Capterra’s own research analyst, Olivia Montgomery, draws the line that matters most for governance: “They’re not decision-making tools, they’re decision-informing tools.”
The Part Nobody’s Advertising
Here is the gap in the research: there is no strong study yet showing that AI-assisted project management tools measurably improve real-world schedule adherence, budget performance, or risk outcomes at scale. What exists is self-reported perception of productivity gains and narrow academic risk-modelling papers, both useful, neither proof.
What is well documented is the risk side. Richard Maltzman at Boston University describes today’s AI tools bluntly: “It’s a clumsy assistant. It cannot be considered 100% trustworthy. There must be human oversight, a human in the loop.” Robert Gordon at American Public University calls the current tool landscape “a little Wild West,” too many overlapping AI features shipped into PM software with no interoperability standard between them. And Capterra’s 2025 trends survey found 41% of PM software buyers now cite AI adoption issues as their single biggest challenge, with 39% admitting their teams lack the AI skills to use what they’ve already bought.
What This Means for PMO Leaders
The practical shift is narrower and more useful than “adopt AI or fall behind,” which is exactly the kind of unearned urgency the 2019 predictions traded on: treat AI features inside your PM tools as an assistant whose output gets checked, not a system whose output gets trusted by default. Build the governance question into the procurement decision, not after the tool is already embedded in how the PMO reports status. And measure whether the tool is actually saving time on the administrative layer it was bought for, rather than assuming it must be working because everyone else says theirs is.
The Question Worth Asking
Project management did change. It just didn’t change the way the 2019 predictions said it would, and the organisations still quoting those predictions as forward-looking insight are the ones least likely to have noticed what actually happened instead.
The honest question for any PMO leader right now isn’t whether AI will transform project management. It already has, quietly, in the parts of the job nobody wrote breathless predictions about. The real question is whether anyone in your organisation has checked what it’s actually doing before trusting it with the parts that matter.
Updated 23 July 2026