ANDREA ORTINO
CHARACTER & CONTROLLER (3Cs)
Companion Control System Redesign
PROJECT - TOWN OF ZOZ
ROLE - DESIGN LEAD
This case study covers the design and iteration of Town of Zoz’s companion control system, a feature that allows players to command a secondary character during combat and exploration. The original manual input model with minimal AI support lacked precision and overwhelmed players, prompting the introduction of more advanced AI assistance.
I led the process from concept to implementation, evolving the system through several iterations: from early AI state logic and a cone-shaped raycast system to a hybrid hold-and-aim mechanic with UI elements. The final design reached a balance between precision and accessibility.
IN A NUTSHELL
CHALLENGES
The companion, Zee, was initially controlled almost entirely through manual inputs (right stick and LT/LB), with minimal AI support that simply mirrored the hero’s target. While functional, this setup overloaded players already managing combat, movement, and abilities with the main character. The lack of clarity in target selection and reaction timing made the system feel unreliable and inconsistent.
The main challenge was to simplify control and improve precision while preserving enough players' agency to coordinate tactics without excessive micromanagement.
APPROACH
ANALYSING THE PROBLEMS
The first step was to gather concrete feedback before proposing any changes. I organised a structured session with the Design and Programming teams, asking participants to document real gameplay scenarios in a shared Miro board, listing expected behaviour versus actual results, inspired by QA methodology. This process eliminated subjective bias and highlighted consistent pain points.
Feedback session notes
Input mapping version 1
IDEATION & ITERATION 1
Based on these findings, I introduced AI behaviour states by reusing the existing enemy AI architecture. This allowed companions to switch between passive, aggressive, and defensive states and use the shared task and plugin system. The goal was to reduce micromanagement while ensuring the companion reacted instantly to player input.
At the slightest right-stick movement, the companion selected a target in front of it using a cone-shaped raycast with centre-weighted priority. Environmental and possessable props were given higher priority during puzzles or mixed encounters to preserve gameplay flow.
While the system felt responsive, playtests revealed it lacked precision: in the heat of combat, players couldn’t always tell who the companion was targeting. The AI assistance additions successfully reduced cognitive load but did not fully resolve clarity or player intent.
ITERATION 2
After brainstorming with the Tech Lead, we decided to trade immediacy for control accuracy. Holding LB activated an aim mode for the main hero, allowing players to manually select a target with the right stick and issue precise orders to the companion. A blue outline highlighted the current target, while environmental props were automatically prioritised when relevant. This redesign made the companion’s intent immediately clear, and all AI behaviour was temporarily overridden during aiming to ensure an instant and predictable response.
New Imput Scheme
Targeting system showcase. I waited a bit before releasing the companion.
OUTCOMES
The new hybrid control system successfully bridged the gap between manual input and AI autonomy. It offered players precision and clarity without overwhelming complexity, improving both combat readability and companion usability. Even though it added one extra input step, playtests showed unanimous improvement. The system felt precise, intuitive, and satisfying to use.
READ MORE
You can read more case studies at the links below. For any questions or further details, you can send me an email at asortino11@gmail.com or a connection request on LinkedIn