Conundrum in scheduling AI application

Time is extremely expensive for some people, while for the others it’s less.

It’s easy to understand why a VC would be interested in an AI startup whose product helps schedule a meeting via emails or phone calls with synthesized human voices using natural language communication. Such a VC will very quickly even pay for the services himself as his time is expensive.

The problem is, people whose time is expensive also tend to be in high-stake businesses. While getting 100 meetings arranged much faster is obviously a plus, getting 1 out of 100 wrong could be disastrous, negating completely the benefit of the time saved.

Less meeting-crammed people, on the other hand, won’t pay for such services. It’s not that their time is not expensive and the app can’t save them money even after them paying. It’s just that they didn’t feel the pain enough to pay for such a tool.

As a result a scheduling AI always runs the conundrum that people who will pay need 100% success rate in the application, while people who are fine with its imperfect performance won’t pay.

I’ve been through this debate with my partners more than 1 year ago. Looking at the AI startup scene today, I feel nothing has changed.

Originally published at http://www.jmyang.com on May 15, 2018.