PA Skills in the AI Era: From Framework to Action

We have spent the last year getting clearer on what AI does to public affairs skills. Which tasks it automates. Which capabilities it amplifies. Which human advantages it makes more valuable by raising the baseline on everything else. We are developing frameworks that are sharper and that will help us navigate the years to come. The BestinBrussels 2026 AI survey — which lands on 25 June — will give us more context on what is being done across the profession in the EU.

But clarity on the diagnosis is not the same as action. And the question I keep hearing — from webinar attendees, from PA leaders and from practitioners thinking seriously about their own development — is not really about the framework so much. It is: so what do I actually do?

That is the question I want to address here.

The knowing-doing gap is real — and it is widening

Most PA professionals have, by now, been exposed to some version of the AI-and-skills story. They have read the reports, attended the sessions, nodded at the frameworks. They have experimented with one or several AI tools (both internally approved ones and then their shadow evening AI work). And then they have gone back to the urgent things on the desk (of which there are always plenty).

All the while the gap between knowing the framework and building a deliberate response to it is widening. And this is key because this is where there is real competitive advantage right now – both as a team and function as for an individual. The practitioners and functions that move from diagnosis to action now will compound their advantage. Those that wait for the profession to reach consensus will spend the next two years catching up.

The action agenda looks different depending on where you sit. But the leadership layer sets the conditions for everyone else — and that is where I want to start.

If you lead a PA function / team

Three decisions matter right now, and none of them requires a large budget or a transformation programme.

The first is a capability audit. Not a performance review — a genuine assessment of where your team sits against both the current skills model and the AI layer. The goal is not to identify deficiencies but to understand development priorities. What does your junior cohort need to stay on a development path that actually builds the judgement the profession needs at senior level? What does your mid-level team need to move from individual AI habit to team-wide workflow integration? Where are the gaps in AI output literacy — the ability to critically evaluate what the tool produces before it leaves your desk? You cannot build a coherent response without first understanding where you are starting from.

The second is getting a system in place. The biggest risk I see in PA functions right now is not that teams are refusing to use AI — it is that they are using it, but chaotically. Individual practitioners have found tools they like. Some workflows have been quietly redesigned. Monitoring has improved. But it has happened by individual initiative, not by design. A system means something different: mapped workflows, clear decisions about which tools are used for what, data governance that actually holds, shared template infrastructure, and explicit standards for output quality. This is not a technology project — it is a management one. AI cannot multiply what is not there. The function that builds the right infrastructure now has something that compounds in value; the one that leaves it to individuals has a collection of personal habits that walks out the door.

The third is embedding this as a team capability, not an individual one. Capability that lives in individuals is fragile. The practitioner who has quietly become the most AI-fluent person on the team is an asset — but they are also a single point of failure. Building genuine team capability means something more deliberate: structured development, shared standards, AI explicitly in hiring criteria, and a culture where the question is not "do you use AI?" but "how well do you use it and how do you help the team improve?" At senior level, this is ultimately a leadership decision about what kind of function you are building — and whether the skills framework you are working to reflects 2026 or 2020.

If you are building your own practice

The individual action is a little simpler, but it is time-sensitive. The window for early-mover advantage in AI fluency is not unlimited. The practitioners who build genuine capability now will compound it into expertise and they will see their value increase on the market as many teams and functions slowly realise the skills they need.

Three things I would suggest. First: pick two tools and build genuine fluency with them. Why two – well I think it is best to pick one that all organisations use (such as Microsoft Pilot) and then say Claude that is streets ahead of the many others. This way you will be expert in something almost everyone has – but you will also be expert on seeing and understanding what else is possible – and this is important. Second: make sure your AI use is is more about workflow than individual prompting. The greater skill here is about automating, templating, standardising than it is about individual prompts. And this is the skills that we will in demand. Third: think about where you sit in the skills hierarchy and consciously invest in the tier above you. The most common career development mistake I see right now is the continued investment in skills where practitioners already have baseline competency, while neglecting the judgment and relational work that is harder to develop and ultimately more decisive.

What comes next

The BestinBrussels 2026 survey will come out on 25 June and will give us more practitioner data to better help us build our PA Skills framework – and also help us understand where we need to move to action. It will be well worth reading when it lands. Mark and I will then hold a webinar about this data – and what it means for the framework and what to do about it – on July 1st 16h CET. More on this soon.

For those who can’t wait: the companion pack from the 7 May webinar is available as a self-assessment tool and development conversation prompt. And if you are thinking about how to run this process inside your function — a capability audit, a system review, or a team capability session — that is a conversation I am happy to have.

The framework exists. The action agenda is now the real work.

What does the knowing-doing gap look like in your function right now? I would be interested to hear where people are — and where the real friction is.