Hi, first of the tutorial in the other answer looks quite good.
Adding on top of that
behavior tree quickstart guide: Behavior Tree Quick Start Guide | Unreal Engine Documentation
eqs quickstart guide: Environment Query System Quick Start | Unreal Engine Documentation
Some faster more in depth view Unreal Engine AI with Behavior Trees | Unreal Engine - YouTube
Some basic introduction, really for complete beginners Introduction to AI with Blueprints - Unreal Engine
And I also found this here from minute 20 to about 38 really interesting for the limits of behavior trees (that’s also why I don’t use them for high level decision making) AI Arborist: Proper Cultivation and Care for Your Behavior Trees - YouTube
And step by step tutorial how to use the build in team system (is C++ only, not possible in blueprints) How do I use the AI Perception Teams? - C++ Programming - Unreal Engine Forums
From my experience you can start out with blueprints and when you start to do crazy workarounds you can think about switching those parts to C++.
For example there is quite some functionality of the AI perception that you can easily access in C++ but not in blueprints (that’s why I also eventually shifted my whole AI perception logic from blueprints to C++). Also you can only implement the build in team system in C++ and if you use the sight perception and have different teams, then imho you should use the build in system (since sight perception prioritizes updates by distance, so actors closer will update more frequently and if you start hitting the maximum number of traces it does on tick (you can increase that in the config files), sight perception will update slower especially for actors further away).
For behavior trees I currently purely use blueprints though.