Phlogiston pulls data from a custom, daily dump of Phabricator data. It was built this way, instead of using Conduit, for [historical reasons I don't remember - maybe Conduit wasn't considered stable enough in 2015, or maybe it lacked some key feature?]. We could rewrite Phlogiston to use Conduit instead.
Would update data in real-time instead of daily. Possibly better support since Phlogiston is the only customer of its dump (afaik) while Conduit is more shared.
Would require a complete rewrite of the loading portion of Phlogiston. Could introduce security issues, since Phlogiston dump has been screened and tested in production against security data leaks.