Posted December 09, 2025 by Kalunae
Hello, my name is Caitlyn and I worked on the Guards in Date the Prince! Over the course of the project, I've worked with designers to get their needs met with the AI, and art to meet their needs with their expected outcome. I didn't have prior experience doing AI or working in a larger team so it was a lot of learning and major changes on my end, but I'm proud of what I accomplished.
Working with everyone separated in their respective teams was a great experience. While most people only communicated through the team's leads, when I did need to communicate with others, it was something I had to get used to since I understand my Blueprints (for the most part). For example, I know how I would toggle a boolean for the guards to see me, but for others, making a debug key or a document explaining how to achieve that same effect was a better. I can also happily say I had 0 merge conflicts during this entire process while working on same blueprints as others! Overall, when I did talk to people, they were very understanding and wanted to learn, and the project went smoothly on my end when it came to requests and communication.
I learned a ton when it came to AI and making it work in different scenarios. There were some wants that were difficult for me to implement easily so I had to do hacks, but that came to bite me later as I had to overhaul a lot as issues kept arising as more and more were added. While it's not anyone's fault, it did take a lot of work, but I learned a lot in the process so now I know for future projects. If I were to change something, I probably should have just worked on simple things in the beginning, then the more complex since design was still designing. That way, I wouldn't burnout early and over things that may not make it to the final version.
To put it simply, a bunch of AI. I had to make 4 different behavior tree that still implemented a lot of similarities so trying to get that to work without full separation was a huge part of this. I created a patrolling guard that follows their patrol route like a good soldier; a sleeping guard who stays in one spot but dozes off, temporarily shutting off their perception; a dog that alerts guards within hearing range of an intruder; and a watchtower that swivels and checks for something not right in its vicinity. In addition to this, I also made the patrol routes the guards follow with different options of patrolling (Turn Around, Loop Around, Stand Still, Rotate) so there's some variety in the routes. Lastly, I just implemented the health and death events on the player and visualized the sight cones on the guards.
In the team devlogs, I tried detailing more about the bugs and what was fixed in that week on the video. The most game-breaking bug that was fixed was the sleeping guards who had no reason being so difficult. For whatever reason, in that current version, they could see you a mile away! And the function that took the most of my time was actually the alert function. For whatever reason, I thought the code HAD TO directly affect the other so they'll know when to attack or go back. I know now just to make a mediator blueprint actor and interface, but when I first tried to approach it, I tried through components since patrolling was a component. It wasn't my brightest moment, especially since in the end, the way I implemented it was just through a MakeNoise node in the behavior tree.
Overall, I accomplished a lot when it came to AI and achieving results based on a design. Even if it's not the greatest AI of all time, I know more than I did 4 months ago so I call that a win.
Thank you for reading this and checking out the game! Have fun with the Prince!