From Speedsolving.com Wiki

KALL is a last layer subset in which you orient the edges and permute the corners at the same time. It only has 15 cases including mirrors, making it easy to learn. It can be useful for people who want to learn an effective last layer subset with out learning many algorithms. It is very similar to OLLCP-A, but it is used as a method, and not as part of method. For example, you get to the F2L-1 cube state however you want. Than, you either use WV, or insert the pair, and than OCLL. Next, you use KALL to get the cube to EPLL, the final step. It is easy to learn as it has few algorithms, and those algs are really easy. Link to alg here: OLLCP-A