At Spaak, we see public affairs teams adopt AI every day. As with all implementation processes, the outcomes are not always the same. Some end better than others, and the pace varies a lot. After 150 implementations, we know that not all AI adoption efforts are born equal. Below is what we have learned.

At Spaak, we have gathered a team of seasoned public affairs professionals who have seen the world of public affairs pre-AI, and who are all now set on the same vision. AI will radically change how public affairs works, and it is crucial that we lead the process with diligence and high ambitions. Our Brussels-based Customer Success Manager, Lukas Schmuck-Demange, joined Spaak before the summer of 2026 from a background in both a consultancy and a highly regulated organisation. Since joining, he has run onboardings across in-house enterprise teams, associations and consultancies. We asked him what he has already observed differentiates the pace and the level of adoption, and what to expect for the future.

Before anyone sees an AI platform

The process rarely starts with logging into an AI platform. It starts with a strategic plan, translated into a more or less detailed vision of how implementing AI will change existing work processes.

Having your team aligned internally on what to achieve with the AI implementation, as specifically as possible, makes it a lot easier to work towards an end-to-end implementation.

Being specific can be really hard with AI, though.

Firstly, because the capabilities of AI change every six months. New models come out. The harness around the models gets more integrated into existing systems and can, to a higher and higher degree, perform full workflows with less human input.

Secondly, because processes are often not very well structured today, either because a sudden change in the political landscape breaks everything in a busy period, or because organisational structures change.

Often, we go in and help the teams we work with outline the plan and the vision, practising some elements of classical change management. But the plan needs to sit on top of a shared vision. Even if AI changes again soon, and is suddenly able to do much more or work in a different way, AI will only have a larger impact from here. Getting good at using it is now a priority for the team, and the plan is how you get started.

Onboarding an AI like a new colleague

We run onboarding sessions with all users, explaining the system and helping them get it set up.

But AI adoption and onboarding sessions are rarely the only to-do in a given week. They are often one more task on an already very long to-do list. It is an investment that will give you a shorter to-do list soon.

In the transition period where the AI takes over a set of workflows, it is important that no task ends up between two chairs, where neither the team nor the AI does it.

AI cannot do everything from day one. It is more like a new colleague who knows everything about the outside world, but nothing about your internal processes. When you tell the AI how you work for the first time, something will almost certainly be misunderstood. A few rounds of calibration will be needed, and that is part of all our implementation phases. Just like onboarding a new colleague.

Internal adoption leaders

Across more than 150 rollouts, we’ve found that every team that implements AI at a high pace succeeds with two things:

• The team that has one or more "superusers", meaning the people who put Spaak into their everyday work in the first week

• The superusers teach the users who adopt more slowly

“So the first thing I do, between the initial onboarding and our next check-in, is find the superusers and work out how they can lead the adoption across the rest of the team. Most often, their own colleague is far more convincing and efficient than I am, because they speak the same internal lingo. I love when the superusers start teaching each other. That is when I know it worked.”

In the first weeks, finding and enabling the superusers matters more to Lukas than the usage numbers. Once the superusers are in place, usage picks up on its own, and people go from a first look at the platform to opening it as the first thing in the morning. “Seeing how Spaak changes the way a team works is what gives this job meaning for me.”

Re-adopting AI as AI evolves

Three months in, Lukas’ read is that the shift to AI in public affairs teams is no longer a question of if, only of how fast each team gets there. And the way you use AI today is not the way you will use it in six months, and probably not in two years either. New models arrive, the tasks the platform can take over grow, and a workflow you calibrated in the spring may look different by the autumn.

So plan to adopt more than once. Every few months, it is worth asking two questions again. Which tasks can the AI take over now that it could not six months ago? And which superusers have found a new way of working that the rest of the team should copy? The second round is a lot quicker than the first. The team already has a plan, it already has its superusers, and it already knows that a new AI capability needs a few rounds of calibration before it works. That is the same three steps, just with a shorter runway.