One thing it’s bad at (ironically because it was made in London) is the data model for Tube stations, it seems to think a Tube station is a point, whereas they often sprawl underground over large distances. It doesn’t have much of a concept that you can be walking for 10 minutes to connect between lines, and will make dodgy suggestions on that basis.
I don’t know how they undercut TfL on the price but I presume they’re doing something tricky with redistributing monthly or annual passes.
They still have consistent problems with randomly charging you random amounts on top of the weekly cost. It's a normal MasterCard so you can check it on the TfL website, but I've had to work with them a few times to get refunded from random extra charges (usually 0.50 to sometimes 7 quid) which they can't explain why they happen.
It's at its best in London (where, IIRC, the dev team is based): there, it's a life-saver. It's also effective in other first-rank anglophone world cities, such as New York. I've found its coverage can be patchy/incomplete in some cities, presumably because the data sources they aggregate are incomplete or inaccurate: at its worst, it degrades gracefully to the state of an offline maps app with a public transit route-finder.
It has been around for some years and they used to integrate with Lyft: I think the idea is to tie into transit pass apps as a way to monetize their service.