The prerequisite for an app is that there are clear yes/no answers for every decision. The reasons this wouldn't work with the current set of regs are:
1. They are not consistent, complete or unambiguous, therefore there's plenty of scope for the same question arriving at multiple different answers.
2. They are sometimes unclear in both intent and detail, so good luck with expressing each sub-reg as a clear statement.
3. Practically no rule set, unless it's extremely simple, can cater for all situations, so the common CASA fallback answer (sometimes justified, sometimes not) of 'it depends on the context' will further cloud the issue.
Fundamentally, to make something like this feasible you would need to first get the razor gang onto the current schemozzle and strip it back to the most basic possible state, then there might be a chance. The problem is forcing the CASR authors to state simply and clearly what they mean by each bit and how it all fits together - again, good luck with that. It would be an easy thought experiment to pick one of the new regs (e.g. Part 135) and just try it - work through from the start and see if you can write out a decision tree that would provide suitable clarity. I'd like to be proven wrong, for sure - I just don't understand how anyone in a position to influence the production of the so-called six pack can put their hand on their heart and say it's a success.
My vote to at least try to cut through the crap is to mandate plain English guides for every Part, publish every answer to anyone from the new CASA Guidance Centre online for all to see, and to have a set of FAQs for each Part that is continually updated.