Creating self-learning tactical collective arttifical intelligence

Is it real to create tactical collective arttifical intelligence with self-learning using blueprints and behaviour tree?