If you want to see where a small or mid-cap tech company is headed to, take a look at their “Jobs” page. Most often than not, it offers a lot of clues on the kind of projects they’re working on, and the technologies that they prefer.
To me, ZenDesk is one of the few companies that the AI revolution should immediately affect, yet they don’t seem to be moving a needle on that front, at least from the surface. I don’t see a single post that mentions Caffe, TensorFlow, Torch or neural networks.
Imagine, neural networks analyzing each customer support ticket with the corresponding response, and eventually proposing canned response to use for new incoming tickets (a la Gmail Smart Reply). Or better yet, detecting frequently touched topics and (a) clustering them as frequently asked questions, (b) forming new canned responses from scratch.
This is not science-fiction, it is feasible, even with today’s technology. Gmail’s Smart Reply, albeit it’s short, is a testament given its non-domain-specific nature. Plus, processing text is much easier compared to images. This year’s results of the one of the toughest AI challenges, ImageNet 2017 are in, and the error rates fell to 2.2% even with such complex categorization tasks. That is down from 16% we saw in 2012 when convolutional neural networks first came out.
Automating customer support, or making it more efficient/pleasant by helping existing staff with well-suited canned responses is one of the areas that may cut many jobs. It is one area where we will see the impact of AI well before the self-driving cars.
ZenDesk has been doing okay since its IPO, but it definitely needs to gear things up. At the end of the day, if ZenDesk doesn’t act proactively here, somebody else will step up; Like SalesForce’s Desk (if it’s not buried in corporate bureaucracy yet!) or FreshDesk.