"Respondable analyzes your messages as you write them, predicts how likely they are to get a response, and then suggests ways you can improve them. If your subject line it too terse, for instance, or the email’s tone seems rude, it will tell you. [..] The startup Crystal scrapes publicly available data about the people you send email to and gives you tips on what style of email you they might prefer based on their personality. [..] The company also relies on public data sets to help train its algorithms. Specifically, Moore says, the company drew upon movie reviews to help its AI learn the difference between text with a positive sentiment and text with a negative sentiment. [..] in many cases, deep learning algorithms are hard to work with because it’s hard to tell why they produce the results they do. “We focused on algorithms that are explainable,” [..] responses drop-off significantly for emails that use language above about a sixth grade reading level, but emails that are too simple tend to be ignored as well. A 3rd grade reading level is optimal. He was also surprised that neutral emails tend to receive fewer responses than positive or negative emails. In short, being too negative is bad, but being slightly negative is better than being boring."