After calculating the weighted sum, the result is passed through an activation function (like Sigmoid, ReLU, or tanh) to introduce non-linearity. This helps neural networks to learn complex patterns.