Can AI replace half of what a PMO does?

It is a fair question, and the honest answer is that AI can take over a real portion of what many PMOs spend their time on today. A lot of PMO work is collecting information from different places, normalising it into a common format, and surfacing where the numbers do not add up. That is assembly and pattern-spotting, and it is exactly the kind of work that AI is good at and getting better at quickly.

If that is the half of the PMO that goes, it should go, and most delivery teams would cheer. The hours spent chasing updates, reconciling spreadsheets and reformatting the same data for three different audiences were never where the value lived. Automating that frees the function from the administration that gave it a bad name in the first place.

What it cannot replace is the half that was always the point. Judgement about what a pattern means and what to do about it. The credibility to walk into a room and tell a sponsor that their programme is not as green as they think. The relationships that let a PMO get the real story rather than the reported one. AI can tell you that a workstream's dates keep slipping while its commentary stays optimistic. It cannot have the conversation that finds out why, and it cannot carry the consequence of the call that follows.

So the risk is not that AI hollows out the PMO. It is that organisations automate the easy half and assume the function is therefore cheaper, while quietly losing the hard half because no one protected it. If you take away the assembly work and do not reinvest in the judgement work, you do not get a leaner PMO. You get a faster way to produce reports that still nobody acts on.

The version I would build leans into this deliberately. Let the tooling do the gathering and the first pass of challenge, and raise the bar for what the people do. More time on the difficult conversations, the cross-portfolio calls, the assurance that something reported actually matches reality. The PMO gets smaller and considerably more valuable at the same time.

For an executive, the trap to avoid is treating this as a cost story. The prize is not a cheaper PMO. It is a PMO that spends its time on the decisions that move the portfolio, because a machine is finally doing the part that was never worth a person's day.