What is the intuition behind solving Ax = b using projections?

We project b onto column space of A to find closest solvable solution.

If b is not in column space of A, no exact solution exists. So we find the closest vector in column space by orthogonal projection, leading to least squares solution.