Slightly fewer, because we used the last 3 interview slots each day to have another track re-interview startups we were on the fence about. That was one of our new techniques for avoiding mistakes. IIRC there were 20 slots both times, so we therefore invited 17/20 as many startups as last time.
I assume by decreasing the amount of startups interviewed in order to re-interview a few of the startups, you guys believe that you make more mistakes in the interview process than in the application review process. That is kind of surprising.