I have been doing this a long time and, frankly, not much has changed in the last 30 years. (Agile vs Waterfall, in its various flavours, cycling in and out of fashion is about it.)

Most delivery environments still feel clunky, including the mature ones. We digitised the admin and called it progress:

• Gantt charts became digital
• RAID logs moved to SharePoint
• Status reports became PowerPoint slides
• Stand-ups replaced weekly meetings

The core problem stayed exactly where it was. Teams still spend hours updating plans, chasing status and reconciling data. That is maintenance, and it eats the time that should be going on delivery.

AI hasn’t solved it yet

Most organisations use AI to draft reports, summarise meetings and generate templates. That saves a bit of time on the outputs. The system underneath is untouched.

What agentic AI actually changes

Agentic AI is a different operating model. It monitors data, makes decisions, takes action and adapts as things move, working towards an outcome without waiting to be told.

What that looks like in practice:

• Task orchestration. Work reassigns based on workload and deadlines.
• Risk management. Risks get picked up early, with mitigation suggested before anyone has to escalate.
• Reporting. Reports generate from live data, with no manual effort.
• Self-adjusting plans. Schedules update as dependencies shift.
• Resource allocation. Teams balance based on skills and availability.
• Portfolio visibility. Multiple projects monitored at once, flagging critical-path risks across programmes.

Coordination becomes orchestration

Agentic PM means orchestrating outcomes through systems that can act on their own.

Worth sitting with for a minute:

• What if plans weren’t maintained by hand?
• What if risks surfaced on their own?
• What if nobody chased updates?
• What if governance ran off live data?

What happens to the project manager

The job moves towards the part that needs a person. Understanding the problem, making the hard calls, and handling the people. The donkey work goes.

Final thought

Don’t spend money optimising a process you intend to replace.