AI has quickly become one of the biggest conversations in the PMO profession and the possibilities are substantial: AI can interrogate large quantities of portfolio information, identify patterns, surface emerging risks, challenge assumptions, reduce administration and make knowledge significantly easier to access.
But the capability of the technology doesn't automatically increase the capability of the teams using it. Before asking what can AI do for our PMO?, you need to know:
Are our processes working?
Can we trust our data?
Are our people ready to work differently?
Is the technology we're introducing actually solving the right problem?
Get those foundations right and technology can amplify good project and portfolio management.
Get them wrong and even the best PPM platform or AI capability will struggle to deliver its potential.
When organisations struggle with reporting, project visibility or portfolio oversight, the response if often to look for a better platform. And sometimes that's absolutely right.
A modern portfolio, programme and project management platform can create consistency across large and complex portfolios, bring information together, reduce manual reporting and give leaders significantly better visibility.
But software alone can't fix unclear governance.
A dashboard can't make inaccurate data trustworthy.
Automation won't resolve a process nobody properly understands.
And a new platform won't automatically persuade people to abandon ways of working they've relied upon for years.
That's why selecting PPM technology should begin with the problem rather than the product.
What are you actually trying to improve? Once that's clear, technology has a much better chance of becoming the enabler rather than an expensive attempt to fix the symptoms.
This becomes even more important when AI enters the picture.
If portfolio information is duplicated, inconsistent, incomplete or out of date, AI doesn't inherently understand that.
It processes the information it's given.
In fact, one of the risks is that AI can make weak information look considerably more convincing.
For PMOs considering AI at organisational or portfolio level, data readiness therefore needs to come first, which doesn't necessarily mean embarking on an enormous data transformation programme. Instead, start with practical questions:
Those are valuable questions regardless of whether AI is involved.
Consistent portfolio information improves reporting, strengthens governance and supports better decision-making.
There's another issue hidden inside the data conversation. People.
Data doesn't become consistent because somebody has created a standard or implemented a new platform. It becomes consistent because people consistently follow the processes that create and maintain it.
Imagine introducing a central PPM platform designed to create one version of the truth where the processes are defined, the governance makes sense and everyone receives training.
Then people return to their day jobs.
One project manager maintains the spreadsheet they've trusted for years, another keeps a separate offline document. Someone updates the central system only when the monthly report is due, and another recreates information elsewhere because that's how their team has always done it. Before long, the organisation has a central system and all the old processes running alongside it.
The technology hasn't necessarily failed, it's just the change hasn't stuck.
Good change management shouldn't begin at go-live.
Engage the people doing the work early and understand which processes frustrate them. Identify where duplication happens and ask why existing workarounds have developed. Explain why consistent information matters and, importantly, show people what's in it for them.
"We've introduced a new reporting process" isn't particularly compelling.
"Capture this information once and you won't need to recreate it for three different reports" is much more meaningful.
That becomes even more important with AI.
People will understandably have questions about trust, security, accountability and their own roles. Those aren't objections to overcome. They're legitimate questions that need clear answers if adoption is going to happen.
Implementation isn't successful because a system went live it's successful when the new process simply becomes the way people work.
Once the foundations are becoming reliable, AI gets much more interesting.
But there's another temptation worth resisting: Adding AI simply because people expect AI.
The test should be much simpler.
Does it make someone's job easier?
Does it surface something valuable?
Does it remove unnecessary effort?
Does it help someone make a better decision?
If the answer is no, it probably doesn't need to be there. And that's the philosophy behind Verto Intelligence.
Rather than applying AI everywhere, we've focused on practical uses such as one-click data insights and specialist AI agents designed around recognisable tasks.
An agent should have a clear job, an insight should help someone understand something, AI should earn its place in the user's workflow and for organisations beginning their own AI journey, the same principle applies.
Start narrow.
Choose a trustworthy dataset.
Choose a real problem.
Choose an outcome you can recognise.
You might want to improve the quality of risk information, or identify gaps in project updates, or surface patterns across multiple projects, or allow leadership to interrogate current portfolio health without waiting for another reporting cycle.
Then test, learn, improve and expand.
That creates evidence, exposes weaknesses in a controlled environment and, crucially, builds confidence.
Imagine a PMO where portfolio information is maintained as part of everyday delivery rather than reconstructed for reporting.
The project teams aren't repeatedly asked for the same information and governance determines who and what can access particular data. AI then helps interrogate portfolio information continuously, which means that potential risks and gaps are surfaced earlier. Ultimately, leaders can explore what's happening across delivery without waiting for the next reporting cycle and PMO professionals spend less time assembling information and more time understanding what it means.
That's the point at which AI stops being simply a personal productivity tool and becomes part of organisational capability.
And there's an important consequence - The more information AI can process, the more important good governance becomes.
→ The faster insight can be produced, the more important human judgement becomes.
→ The easier it is to create a report, the more important it becomes to ask whether it's the right report.
That changes where PMO professionals create value:
Less time collecting.
Less time manually consolidating.
Less time working out what happened three weeks ago.
More time interpreting.
More time challenging.
More time advising leadership.
More time improving delivery.
That's a more strategic PMO.
So before investing in another platform or embarking on the next AI initiative, we'd suggest asking four questions:
Are they clear, proportionate and consistently understood?
Is it current, consistent, governed and appropriately accessible?
Do they understand the change, why it's happening and what it means for the way they work?
Can we clearly explain what it makes easier, faster or better?
If the answer to all four is yes, technology has a genuine opportunity to amplify good PMO practice. If not, the work that needs to happen first has probably just become much clearer.
This is the conversation we'll be taking to House of PMO Conference in Edinburgh on 16 and 17 November, where Pete Cox, Natasha Osborne and Simon Sharp will explore data quality, organisational change, PPM technology and practical AI with PMO professionals from across sectors.
Better processes. Better data. Better adoption. Better technology. Better decisions.