As you've mentioned, India is a leader in generic drugs but even here; middle income population are largely not aware of the availability of cheap generic equivalents for branded medicine prescribed by the doctor as most clinics and hospitals coerce the patients into buying medicine from their own store in-spite of govt. taking various measure to promote generic drugs (e.g. separate unbranded generic drug stores - Jan Aushadhi, considering amendments to the Drugs and Cosmetic Rules for mandatory generic medicine prescription).
Fun fact, generic medicine on prescription is mandatory since 2016 in India according to Medical Council of India's ethics code but it's not enforced. So solving this need gap could be very useful even here.
As for the solution you've suggested, handwriting recognition is one of the first exercise almost every Machine Learning enthusiast practices; so identifying data from prescription thanks to better cameras and compute power in smartphones should be no-brainer. But, organising drug information is the key; it can be procured from the govt. but doing it globally would be the pain point of building a solution for this problem.