Developed a Behaviour Tree system supported by custom State Trees to handle roaming, investigation, and chase behaviours. Early iterations suffered from pathfinding issues due to the enemy’s scale, causing clipping on corners, stalled movement, and inconsistent state transitions. To resolve this, navigation settings and collision were refined, and State Trees were introduced to stabilise behaviour flow. This resulted in a more reliable and responsive AI system that maintains pressure on the player without breaking immersion.

Designed an interaction-driven progression system using inventory items, keycards, and locked doors to control player flow. This system was used to gate access to new areas, allowing tension to build gradually as players move deeper into the environment. Level structure was simplified over time to support clearer navigation and pacing. The result is a more controlled progression system that reinforces tension through structured exploration rather than open-ended wandering.

Implemented a mix of scripted and system-driven encounters to create tension throughout the experience. Early scare events lacked context within the environment, leading to a shift toward more player-driven interactions. The AI was supported by a teleport system using predefined spawn points, allowing it to reposition closer to the player and maintain pressure. Encounters were further enhanced by introducing distraction mechanics, where players could use throwable objects to manipulate enemy behaviour and create escape opportunities.

Focused on ensuring all core systems function cohesively without breaking under gameplay conditions. Initial development revealed issues with AI state management, including getting stuck or failing to transition correctly. Moving from a purely Behaviour Tree setup to a hybrid system using State Trees significantly improved consistency. This focus on stability ensured that AI, interaction, and progression systems work together to deliver a smooth and reliable player experience.