This paper presents the results of randomized controlled trial conducted to evaluate a Grade 8 after-school mathematics intervention in Cape Town which aimed to fill knowledge gaps using computer assisted learning. The programme employed student coaches to facilitate classes in which Khan Academy resources were used to teach basic numeracy. Large gains of 0.321 standard deviations were observed on basic numeracy outcomes for learners who were selected to be on the programme. Similarly, learners in treatment also scored 0.246 standard deviations higher on core Grade 8 curriculum questions at endline. The improvements in mathematics outcomes were evident for learners throughout the distribution.