Reforged Labs
Reforged Labs provides game developers with AI so that they can create quests and characters that react incredibly realistically to player choices.
Created using Figma
Table of contents:
Background
Task
Execution
Conclusion
Background
Video games are the ultimate form of immersion. World design, character design, gameplay, and more are all carefully curated to draw players in as much as they can. However, this fragile web can come crashing down at a moment’s notice when something seems off. Sometimes they might be bugs, but oftentimes it’s a limitation of the game itself — like when low-level bandits try to mug the same character who has slayed dragons.
Context
Photo of some dialogue from the dialogue rich Baldur’s Gate 3
Reforged Labs uses AI to create characters and quests that adapt to player choices. In doing so, games can feel more immersive while saving writers and developers precious time that would have been otherwise needed to flesh out every possible course of action.
Research
I interviewed a few indie game developers who had extensive experience writing characters, worlds, and the stories that transport players to them. Although I could only interview a few people, I took an hour each to truly ensure I was getting as much as I could from them. One of the interviewees even had experience creating their own reactive characters using chatbots. Some interesting quotes included:
Since I conducted these interviews over Zoom, I was able to get a transcript which made it very easy to review afterwards. I noted down notable quotes in a spreadsheet, adding context and my own thoughts. Not only did this make it easier for me to analyze my notes, but in the future it would also make it easier for a team to quickly catchup with the insights that I gleaned.
While I only interviewed indie game developers, I had a feeling that the main source of revenue for Reforged Labs would be from game studios. Although I wasn’t able to get a hold of any writers at bigger game studios, I used some of the insights from the interviews I did have to extrapolate what they might be looking for. Here are a couple of personas detailing both parties’ needs and struggles:
Task
dfsadf
Beardo giving his thoughts on my project
Surrounding his house with trash in response
Execution
Throughout the project we were constantly iterating on the designs. However, one core theme stuck with us: game visuals. The placement of the text, even whether the text popped up on the screen versus came up one letter at a time was important to achieving that game-like feel.
Documentation was quite important not only for the client’s understanding but also for the developer. We had base exercises/interactions that would be used in many of the lessons, so we categorized where those would be based on the order of the lessons as well as grouping them by exercise. Furthermore, I held a series of alignment meetings to explain how the patient would progress through the game.
Wireframes
Documentation
The Game
So, how did we translate Animal Crossing to mental health? First, we picked out the pieces that we felt like could solve some of the pain points the patient felt. For the relaxing environment, I suggested a garden as it could be a place to meditate. In addition, the outdoors and greenery would be relaxing. Then, the question of the more human side came in. Did we want to have the character looking at you? Did we give the patient their own person who the character talked to? For many of these considerations, I discussed with the clients to make sure they understood the reasoning behind our suggestions.
A series of moodboards that we used to align with the clients and give suggestions
Before and after: dialogue
Documentation for which interactions go with which lessons
Conclusion
Wireframes
Although we had some setbacks in the end, the clients were happy with the final result. We had gamified a few of the lessons so that playing the game would actually reflect the learning objective. In addition, the art assets were high quality and would help immerse the patient in a relaxing island adventure. The end product comprised 62 screens and dozens of art assets. Everything, from signup to the exercises, would be there for the clients to pass on to their developers, and with the documentation they would have little problem figuring it out.
Reflections
The main gap that was highlighted during this project was my visual design skills. While the basic wireframe was helpful for Brandon to go off of, my high-fidelity wireframes would often have to be polished by him.
As the team lead I also should have allocated more time towards the creation of the art assets as it became a bottleneck towards the end of the project.
Proposed KPIs to Track
% change of Patients that finish the program:
Measuring how much more effective the new program is than conventional approaches
Finishing the program is one of the key objectives for Hollo
% of Patients who return to therapy within 6 months:
Finishing the program is pointless if it has had no effect on them and they still require therapy after
Hollo communicates with therapists so this information would be available
However, just because patients still require therapy does not mean the app has had no effect
