If you're building a tool or an app and your whole team is three people, and you're doing this on the side while you're teaching or whatever, the idea of running a three-year randomized controlled study doesn't make sense," he said. "We need to show how that can be accelerated.