Every project management tool now has an “AI” badge. Asana has AI, Monday has AI, ClickUp has AI, Notion has AI. The marketing suggests these features will revolutionise how your agency manages projects.
The reality is more nuanced. Some AI PM features are genuinely useful. Others are gimmicks that add complexity without saving time. Here is what is worth your attention.
What works
Automated status updates. The most immediately useful AI feature in project management. Instead of someone manually compiling a status update from task completion data, the tool generates one automatically. You review, adjust, and send. Saves 15-30 minutes per project per week.
Task summarisation. For complex projects with hundreds of subtasks, AI can summarise progress at a glance: what is on track, what is behind, and what needs attention. This is particularly useful for account managers who oversee multiple projects.
Meeting action items to tasks. Connect your meeting transcription tool to your PM tool and action items from client calls become tasks automatically, assigned to the right person with the right context. This closes the gap between “we agreed to do this” and “someone is actually doing it.”
Workload prediction. AI analyses your team’s current task load, historical completion rates, and upcoming deadlines to flag capacity issues before they become crises. Not perfect, but better than the spreadsheet most agencies use. For more on this, see our guide on AI for resource planning and capacity management.
Natural language queries. Instead of building complex filters, ask “what tasks are overdue across all projects this week?” and get an answer. Useful for founders and account directors who need quick visibility without learning the tool’s query syntax.
What does not work (yet)
AI-generated project plans. Every tool offers this, and every output is too generic to use. A project plan for a website build is not the same across agencies, clients, or contexts. Your templates, refined over years, are better than what AI generates from a brief description.
Automated resource allocation. The idea is sound: AI looks at team skills, availability, and project requirements, then suggests assignments. In practice, it misses the context that matters most: who works well with which client, who needs development on a particular skill, and who is already stretched thin in ways the tool cannot see.
Predictive risk analysis. Some tools claim to predict project risks based on patterns. The data set within a single agency is too small for this to be reliable. The risks that matter most (scope creep from a particular client, a team member burning out, a third-party dependency failing) are not the kind of risks that statistical patterns catch.
The honest assessment
AI in project management is most useful at the edges: automating the reporting and communication tasks that surround project work. It is least useful at the core: the actual planning, resource allocation, and risk management that good project managers do.
The best approach for agencies right now:
- Use AI for status updates and communication automation. This saves real time with minimal risk.
- Use AI for meeting-to-task conversion. This closes a genuine gap in most agency workflows.
- Ignore AI-generated project plans. Your templates are better.
- Keep human judgement at the centre. For resource allocation, risk management, and client communication, experienced project managers outperform any AI feature currently available.
The PM tools that add the most value are the ones that automate the admin around project management, not the ones that try to replace the thinking. The same principle applies across agency automation: automate the repetitive work, keep human judgement where it counts.
This is part of Tool Drop, a series reviewing AI tools and approaches through an agency lens. Subscribe to the newsletter to get new articles weekly.
