Last Thursday I had the pleasure of speaking at the House of PMO Conference in London. If you haven't come across House of PMO before, they are the UK's professional body for PMO practitioners and they do a brilliant job of bringing the community together, challenging thinking and giving practitioners a proper forum to share what's actually working on the ground. The energy in the room was great, and it was one of those events where the conversations in the breaks are just as valuable as the sessions themselves.
My session was titled "AI Meets PMO: Cutting through complexity to drive public value." I want to use this post to pull out the things that matter most, because a 45-minute slot goes fast and I know not everything lands in the moment.
The problem we're actually trying to solve
Before any conversation about AI makes sense, you have to be clear on the underlying problem. In most public sector organisations, programme and project management is genuinely hard not because people aren't capable, but because the conditions they're working in are stacked against them.
Think about what the average PMO is dealing with: progress, spend and benefits being tracked inconsistently across teams; management information that takes days to pull together and still doesn't quite reflect reality; and data sitting in Word documents, Excel spreadsheets, SharePoint folders, finance systems and email chains, with no single version of the truth anywhere. On top of that, a significant proportion of the people managing projects never set out to be project managers. They're brilliant subject matter experts who found themselves responsible for delivery because someone had to be.
This is the environment AI has to operate in. That context matters, because it shapes what good AI looks like in practice.
AI has to earn its place
I want to be direct about something I said on Thursday, because I think it's the most important point in the whole session.
AI must provide a genuine benefit to the end user. Not to the vendor. Not to the marketing department. But to the person sitting at their desk trying to make sense of a complex programme and make a decent decision before the next board meeting.
There is a lot of AI washing happening in the market right now. Products are slapping an AI logo on features that amount to very little. That is not what we built and it is not what the public sector deserves. If an AI feature does not meaningfully change the experience of the person using it, it should not be there.
The second non-negotiable is security. Your data and the information being processed by AI must be handled within your information governance frameworks. Full stop. This is not a nice-to-have. It's the foundation that everything else has to be built on.
What AI can genuinely do for PMO teams
With those principles in place, the opportunity is real and it's significant.
AI can analyse every update across every project in a programme in seconds. For a PMO team that would otherwise spend days consolidating status reports, that is a meaningful shift in capacity and quality. It can challenge human assumptions, flag things that look off, and help people who are under pressure make better-informed decisions faster.
One area I feel strongly about is what I'd call the blank page problem. A lot of people who are managing projects, especially accidental project managers, find it genuinely difficult to know where to start when writing a risk entry, a status update or a lessons learned log. AI can help with that. It can take unstructured data, like notes from a meeting or a spreadsheet that doesn't map neatly to your system, and do most of the heavy lifting to get it into the right shape. That's not replacing judgement. That's giving someone a solid starting point they can then own and refine.
The other concept I introduced on Thursday was Verto Intelligence Partners, which are agentic AI tools designed to bring specialist expertise directly into your programmes and projects. The idea is that your AI should understand your organisational context, your PPM framework and the lessons your organisation has already learned. A generalist AI that knows nothing about your environment will only get you so far. Specialists that are embedded in your data and your processes are where the real value sits.
Bringing AI to the data, not the other way around
One of the key takeaways I want people to leave with is this: the right model is bringing AI to the data, not extracting data and sending it somewhere else to be processed.
This matters for security, obviously. But it also matters for quality. The more context the AI has, the more useful it is. If you're copying and pasting data out of your systems into a generic AI tool, you're losing the surrounding context that makes the output trustworthy. You're also creating a governance headache you probably don't need.
What I hope you took away
If there were five things I wanted people to walk out with on Thursday, it was these:
- AI has to provide real benefit to the end user, or it's not worth the noise.
- Security and information governance are not optional extras.
- Your AI agents should be specialists, not generalists.
- Bring AI to the data and not the data to the AI.
- And for the many accidental project managers working across the public sector right now, AI is one of the best upskilling opportunities they've ever had access to.
It was a great event. House of PMO consistently creates a space where the conversation feels genuine and the audience is asking the right questions. If you were in the room on Thursday and want to dig into any of this further, or if you couldn't make it and want to see what we've built, contact me and we can arrange to speak. For an idea of what was discussed, you can rewatch the webinar we recorded about this topic, below.
/digigov--ross-talk-3.png?width=720&height=380&name=digigov--ross-talk-3.png)