Before you start adding the AI buzzword to every single tech you're building...ask first What features in my product "require" some form of artificial intelligence? Then search to see what techniques can be used to implement those features.
Maybe those features might not require AI at all. In which case, great!
If they do require AI, it is likely that it require some form of AI that is already so commonplace that you might already know how to do it already. If not, there's probably many tutorials online that can quickly teach you how to implement that AI technique. That's what you need to search for.
Consider the A* algorithm that is commonly used for pathfinding. Pathfinding is important for NPCs in video games and simulations. Pathfinding is considered a technique that traditionally would seem to require intelligence (you're moving around in the world). Ergo, pathfinding would technically be an AI algorithm. Yet, the A* algorithm was already discovered in the 1960s and is today taught in undergraduate Computer Science courses. You would likely need AI to handle pathfinding...but you would probably already know how to implement that type of AI (or at least, be able to quickly find out through the various tutorials online about the A* algorithm).
I define AI as any sufficiently complex algorithm (an admittedly broad definition). Under my definition, AI is all around us. The upshot is that most programmers will write AI, whether they realize it or not, and so there's no real need to think about how to "add" AI to a project. Instead, simply worry about the requirements of your project, and then implement those requirements (pulling in AI techniques if necessary).