I think that depends on how sophisticated a service is. I agree that applying some regular expressions to emails is not enough to make any other company buy you instead of building it themselves. But that holds for almost anything a startup can do.

Startups need to solve difficult problems or they are nothing more than marketing firms trying to create a hype around hot air. Filtering email intelligently, including automatic anonymization of sensitive parts within the email, can be a difficult problem to solve, and I think there are many types of email integration that a startup could tackle based on the approach Matt suggests.