Maxxton opened a new AI hub in Amsterdam in spring 2026 to build AI agents for outdoor hospitality software. The team works on a question that could change outdoor accommodation businesses use that software: what happens when employees no longer have to learn it before they can use it?
Rather than adding another AI feature or chatbot, the team is working towards agents that understand what an employee or guest wants, use the information already available in the PMS and help carry out the work.
The new team brings together different AI disciplines. Nino Foudraine leads the initiative, bringing six years of AI experience and a background in econometrics.
Gijs Gubbels focuses on conversational agents and agentic AI; Martin de Smit works on backend architecture, models and data; and Roelant de Looff focuses on the frontend and, importantly, what can actually be built. Amsterdam gives Maxxton access to a wider pool of specialised technology talent to hire from.
Nino Foudraine believes the change goes beyond adding AI to existing software. “The world is changing in the AI era,” he says, pointing to the rapid development of AI and the move from Software as a Service towards “Agent as a Service”. His examples are already familiar: phones with built-in agents, AI deciding which music to recommend, and eventually booking a holiday by simply telling an agent what you want. “It’s a shift in user behaviour,” he says.
For outdoor hospitality operators, that shift could matter most where software is currently hardest to use: in day-to-day operations. Seasonal employees, field workers and new colleagues often need to learn complex processes and find their way through multiple screens before they can work independently.
Max is our agent designed to change that relationship with the software. The assistant sits alongside the Maxxton platform and helps employees find the information or functionality they need rather than expecting them to know where to look. The result is already visible in onboarding: what traditionally takes around six weeks can now take roughly one hour.
The bigger opportunity is not simply faster training. “Previously, you’ve been clicking,” Nino Foudraine says. “AI makes it conversational.” Instead of learning where a function lives in the system, an employee can start with the problem they need to solve.
Early customer use suggests that this is more than a different interface. At De Krim, field workers were sending around eight questions a day to the back office; once Max was introduced, seven of those eight questions were resolved internally.
That changes the economics of support as much as it changes the user experience. Fewer questions have to move between the park and the back office, experienced employees are interrupted less often, and field workers do not have to wait for someone else to know the answer.
Maxxton is also using those interactions to identify where the software itself needs to improve. In testing with Libéma, questions are being grouped into three categories:
The result is a feedback loop between AI and product development. “Analyse how people work to pinpoint automation opportunities,” is how Nino Foudraine describes the broader approach. The agent is not only answering questions; it goes further by showing Maxxton where users struggle with the product.
The same shift is taking place on the guest side. Ton is our agent designed to support the guest experience from the first search through booking, pre-arrival communication, the stay itself and what happens afterwards.
That means a guest could use a conversational interface to find a suitable stay, receive relevant suggestions such as pet packages or transport, ask questions during the visit and make bookings without having to navigate separate systems. The agent can work with the guest’s existing context across channels such as phone, email and WhatsApp.
For operators, Nino Foudraine sees three forms of value:
Answering questions is only the beginning. The bigger challenge is giving an agent enough access and context to actually perform tasks on behalf of an employee or guest.
That is where the PMS becomes important. Max and Ton live inside the Maxxton platform rather than operating as disconnected AI tools. “It’s complementary, not an ‘extra’,” Nino Foudraine says. The existing Maxxton software remains in place; AI acts as a helper and guide on top of it.
For operators, this also addresses one of the biggest concerns around enterprise AI: data. “All in the same environment, within PMS, data protection, within the ecosystem,” is how Nino Foudraine describes the advantage. A customer can copy information into a third-party AI tool, he points out, but may not know where that data ultimately ends up.
The technology inevitably raises a more difficult question for frontline teams: if an agent can answer questions and perform tasks, what happens to the people doing that work today?
Nino Foudraine compares the transition with the arrival of the steam engine. “Yes, some repetitive jobs will disappear, but new professional roles will emerge in their place.” His argument is that automation does not necessarily remove the need for people; it changes where their time and judgement are most valuable.
That distinction may be particularly important in outdoor hospitality, where operators already rely heavily on seasonal teams and where experienced employees often carry a disproportionate amount of operational knowledge.
Nino Foudraine expects the next two to three years to bring a more fundamental change in how operators interact with their systems. “Operators will move from clicking through software to simply stating a goal and letting an agent carry it out,” with the system validating its own work before returning the result.
The same principle could transform the guest experience. “Guest reporting becomes hyper-personalised almost by default,” because the system already has the context needed to understand the individual guest.
That is ultimately what the Amsterdam investment represents: not another AI feature, but a bet that hospitality software will become less visible to the people using it. For outdoor operators, the question may soon be less about how quickly their teams can learn the software and more about how much of it they still need to see.