It doesn't have to a "big bang" (ie, tackle EVERY query).
You could start with something like:
select FORCE_MATCHING_SIGNATURE, sum(execution_count)
from v$sql
group by FORCE_MATCHING_SIGNATURE
which gives you the highest execution count SQL - you could target them initially to get the biggest "bang for your buck". Convert those to (say) PLSQL to get all the benefits of lower parsing, cursor reuse etc.