For beginning programmers, understanding how to express instructions to the computer and how to organize these instructions can be difficult. We propose using AI to guide students in their first programming steps through reverse code completion. This means that instead of the traditional scenario where editors provide code completion, we suggest that novice students need a starting point such as a template, to get into programming. Our proposal is a new learning strategy based on completing lines of code provided by Artificial Intelligence through a step-by -step guided process. Our experimental results show a decrease in the amount of time it takes a student to solve a programming problem and an increase in the success rate. Furthermore, we have been able to demonstrate that knowledge is acquired through this approach by successfully solving similar problems after receiving AI support. In addition, we compared the overall performance of the students who were part of this study regarding the same course in previous years and we obtained a higher grade average with a faster growth in the learning curve, especially in the first evaluations. Finally, we measured the Zone of Proximal Development and we found that the number of students below the zone decreased over time. In other words, the ability of the students to solve a problem on their own was increased.