I’ve spent the last two years systematically testing AI tools for project management. Not casually. I mean building workflows, running them on real projects, evaluating what actually helps versus what creates more overhead than it saves.
The landscape is crowded, the claims are often exaggerated, and the hype cycle is in full swing. Here’s my honest take on what’s useful, what’s overpromised, and what I actually reach for every day.
The Categories That Matter
There are roughly five areas where AI has meaningful PM applications. They’re not all created equal.
1. Meeting Intelligence: Genuinely Excellent
This is the clearest win. Tools that transcribe, summarize, and extract action items from meetings have saved me a significant amount of time each week. The technology has matured quickly, and the output quality is good enough to use with minimal editing.
The key feature to prioritize is action item extraction, not just summarization, but identifying who owns what and by when. Real talk: the output still needs human review. AI will flag what was said; you need to make the judgment about what actually matters and what was just noise in the conversation. That’s not a flaw. That’s the correct division of labor.
2. Documentation and Status Reports: Very Good
AI is excellent at taking messy notes and producing polished communication. Give it your rough project notes, tell it the audience (executive summary, team update, or client-facing), and it will produce a well-structured draft in about 30 seconds.
I don’t use the output verbatim (I always revise), but the time savings on first drafts are real. What used to take 20-30 minutes now takes 5, with better structure than I’d produce from scratch under time pressure.
3. Risk Identification and Pattern Analysis: Promising, Not There Yet
This is the category with the most exciting potential and the most uneven execution. Several tools claim to identify project risks by analyzing your data. The pattern recognition is genuinely interesting, but the signal-to-noise ratio is still too high to rely on.
What I’ve found more useful: using AI conversationally to pressure-test my risk assessment. Describing a project situation and asking “what am I missing?” generates useful prompts for thinking. It’s not the same as algorithmic risk detection, but it’s practical today.
4. Planning and Scheduling: Mixed Results
This is where I’d urge the most caution. AI-generated project plans look authoritative and are often structurally wrong. They don’t know your team’s velocity, your organization’s constraints, your specific client’s risk tolerance, or the dozen informal dependencies that exist in every real project.
AI planning assistance is useful for structure and completeness checks. “Did I miss any phase?” is a good question for AI. “Generate me a timeline” is usually a waste of time that you’ll then spend correcting.
5. Stakeholder Communication: Genuinely Useful
Writing tailored communications (adjusting tone, level of detail, and framing for different audiences) is something AI handles well. Drafting a difficult message to a client, adjusting a status report for an executive who prefers one-page summaries, reframing a delay as a proactive heads-up rather than a failure: AI is a solid collaborator for all of this.
What I Don’t Recommend (Yet)
- Fully automated AI project managers. Several tools market themselves as AI systems that “manage the project for you.” A project without a human PM accountable for its outcome is a project that’s going to have a bad day at some point. Automation for the administrative work is great. Automation for the judgment work is dangerous.
- AI features bolted onto existing PM platforms. Several established project management platforms have added AI features that are, frankly, gimmicks. Tab completion in your task descriptions is not AI-assisted project management. Be skeptical of feature releases that are more about marketing than utility.
The Bottom Line
AI is a real productivity multiplier for project managers who approach it with clear thinking about what problem they’re trying to solve. Used well, it eliminates a meaningful portion of administrative overhead and sharpens your communication. Used naively, it creates more work and a false sense of coverage.
The PMs who benefit most are the ones who maintain a clear sense of what human judgment is irreplaceable for, and use AI to clear the path for more of that work.
That’s the entire thesis behind ThirtyYearPM. More on that soon.