Describing a variant on a different strand is always confusing.
If you are just talking about the genotype, you have to create the reverse complementary sequence. This means you start reading your initial genotype from right to left and changes each base to it complement (A/T or G/C).
So the reverse complement to AG would CT.
If you take the position of the variant into account, it looks a bit different. Positions are always described on the forward (+) strand. And if you describe an interval, you always starts with the lower position. So e.g. you have a variant AG on the forward strand starting on position 5, the equivalent description on the reverse strand would be TC starting on position 5 as well. You see I just created the complementary sequence and not the reverse complementary.