Welcome to Derry

Before stepping inside, guests were asked a simple question: What are you afraid of? That single input became the fuel for everything that followed. When guests entered the space—a 1960s-inspired child’s bedroom—their fear was generated live, revealed on screen and threaded through audio that immediately began reshaping the room around them.

From there, nothing played out the same way twice. Sound design tightened or dragged. Timing shifted. The atmosphere recalibrated on the fly. Instead of a fixed scare sequence, the experience responded to each guest in real time, making it feel less like an installation and more like something watching them back.

Under the hood, we built a generative system designed to flex. Guest inputs were processed live and mapped to a modular audiovisual framework, allowing the environment to continuously recompose itself rather than trigger a single pre-authored outcome. We weren’t interested in one perfect scare—we wanted a system that could adapt, escalate, and surprise at the pace of a convention floor.

Rather than writing one narrative arc, we created a toolkit of sound, timing, and narrative variables that could be assembled differently for every guest. The result was an experience that behaved like a character: reactive, intentional, and just unpredictable enough to keep people on edge.

The goal was never AI for AI’s sake. It was about building a form of experiential storytelling that could move faster, respond in the moment, and connect more personally than traditional activations allow. By turning each guest’s own fear into the source material, we created horror that felt intimate, unsettling, and impossible to shake off.

Welcome to Derry became a proving ground for how we think about AI inside physical spaces—not as a replacement for creativity, but as a collaborator that helps us build worlds that feel more alive, more human, and, in this case, a lot more terrifying.