We use Selenium for our FDA validation testing.
For those not familiar with FDA and NASA quality control initiatives and testing requirements, suffice it to say that the federal government tends to talk in terms of 'validation and verification testing' in many of its' regulations. There's decades of history behind this term, and the three important takeaways are the following:
First, a project has to have two or more testing strategies, so as not to fall into a bias with a single testing strategy. Second, one of those testing strategies has to be user-centric, commonly known as validation testing. Examples include crash test dummies and end-to-end software acceptance tests. Third, most any other testing strategy can qualify as verification testing.
So, we use Selenium for our FDA validation testing, because it simulates a user walking through the software.
It's big and bloated and slow, and the legacy version runs on Java, and it's brittle and doubles the amount of code we have to write, and is otherwise a total pain in the ass. But it's like folding your parachute before skydiving or having a depth gauge when SCUBA diving or placing bolts when climbing a wall. There are simply some tasks that you don't want to do without safety gear.
And for all its' pain to use, Selenium is like safety gear for building your tech stack. We've migrated across a half a dozen user interfaces, two entirely different back-end languages, three or four data protocols, and countless utilities over the years. And Selenium warns us when we're in danger and it catches us when we fall.
