As a former Flex developer, I must admit that Flex was the best thing that ever happened to web development, but unfortunately the Flash team wasn't strong enough to bring the runtime to mobile with enough quality. Building a RIA in Flex was so incredibly easy, and debugging was easy, and you really could use anything you wanted for a backend (I've used Perl, PHP, and Java). Bidirectional data binding, an excellent widget library, and a lot of help from the default IDE (custom build of Eclipse) allowed me to build software, full stack, with an engineering team of 1 on a weekly release cycle.
I've spent more time debugging HTML5/JS/CSS than building entire applications in Flex. The AdvancedDataGrid component alone was worth the price of admission.
I used to work for Macromedia but I knew the moment Flex was released they'd taken the wrong turn from complimenting web standards to trying to build their own proprietary platform on Flash. To the user it was only marginally better than Java applets.
Flex is still good for some $100-$120/hr consulting gigs on the side, mostly maintaining or changing something for a company who no longer employees a full time flex dev. Each year since Adobe "killed Flex" this hourly rate has gone up while the # of gigs has gone down. I would not, nor recommend, anyone start a Flex app today for a new project.
AngularJS is the closest thing to Flex that I've used, and only recently has tooling started to come close to what Flex had years ago.
JS/CSS is painful in comparison (esp when you consider large corps are still on IE6, 7, or 8), but still the right choice today for new RIAs. One can only hope the death of XP next year gets these corps to to do something...
For a web app (RIA) I would pick Flex or Silverlight any day over the whole HTML mess if they would only run properly in all browsers, desktop and mobile.