I coded a script to extract data from transcripts of an automated customer support chat and enter them into a CRM. Even if customers had to answer quite precisely defined questions (example: birth place and dd.mm.yy) they make all sort of mistakes, which won't be mistakes at all in a normal conversation. Do you have something to handle messy data or does your tool works like copying all it sees in a position to a form field with no further processing? Obviously processing can be done later on and better questions might be asked.
The tool works best for machine generated emails where the structure remains the same from one email to the other.
That being said, the parsing engine is quite resistant to noise. For instance, you can have changes in the footer or some other sections of the email, or have additional forwards and it will be able to get the data without issues.
Regarding cleaning up the type of data itself, we don't do any post processing at the moment besides trimming spaces and removing html tags. We plan to add data types at some point (for instance to convert any text date into a standard date) but we'll be waiting to have a paying customer request for it before it gets implemented :-)