First things first
Spotify filed for its IPO on Wednesday and, not surprisingly, the company's F-1 filing places a heavy emphasis on Spotify's use of artificial intelligence and machine learning to power personalization. In fact, here's the company's leading bullet point in the section describing growth strategies:
We will continue to (i) invest heavily in research and development, (ii) make strategic acquisitions in order to enhance our product capabilities, and (iii) make our offerings more attractive to existing and prospective Users. We will continue to invest in our artificial intelligence and machine learning capabilities to deepen the personalized experience that we offer to all of our Users.
I don't know for sure, but I suspect Spotify's early embrace of deep learning to power machine listening has paid large dividends over the past few years. Basing recommendations and playlists on genre and related artists is tried and true, but the ultimate factor behind why we enjoy music is how it sounds. And only now are we actually able to (1) catalog pretty much every song ever and (2) use machines to analyze how they're constructed.
On the acquisition front, Spotify also notes how some of its acquisitions over the years -- including the Echo Nest one that I covered years ago while at Gigaom -- have bolstered its AI capabilities.
Overall, I'd argue that Spotify is another great example of how AI can actually be applied successfully today, without getting hung up on how it's going to completely transform your company or the entire world. You have data that's amenable to these types of models, you analyze it, and you use the results to improve your customers' experience. At the current moment, especially with the mass-market tools coming online, I think that's a prudent and very achievable approach.
Apple probably shouldn't be paying Google or AWS for cloud storage
It's been pretty widely known that Apple has been using AWS for iCloud storage, and this week news broke that it's also now using the Google Cloud. That's probably a bad idea, for some of the reasons that I pointed out in the last newsletter about Dropbox's IPO and move from AWS to its own infrastructure. Even if cloud storage is cheaper, easier and not really a competitive advantage, it's hard to see how lining your blood rivals' pockets is a good idea in the long run.
This is especially true as consumer tech moves into the era of smart devices, which are part of a virtuous cycle wherein devices, data and cloud services all feed off each other. It would be a different story if Google and AWS were just selling cloud computing infrastructure, but they're not. And by continuing to support those business lines, Apple is indirectly helping improve Alexa and Google Home, and also whatever third-party products emerge from the AI APIs AWS and Google are selling to developers.
I've really come around on Apple's focus on making great hardware and putting user privacy first, but looking at it coldly it's difficult to see how Apple will continue to compete in the smart-home era if it doesn't start to own its entire supply chain.
If you're not thinking about GDPR, now's a good time to start
I'm keeping this brief, but the bottom line is that aside from mandating where data is stored and how breaches must be handled, the EU's GDPR also places regulations on how data can be used. Companies and applications that like to use data in inventive ways probably want to get their privacy policies and user consent methods in order. Here are a couple articles talking about this issue:
- The GDPR: An artificial intelligence killer? (Datanami)
- How GDPR will change the way you develop (Smashing Magazine)
Two more things
- On securing the Kubernetes dashboard (Heptio): Remember how Tesla got cryptojacked via an unsecured Kubernetes dashboard? Joe Beda of Heptio (and a Kubernetes creator) explains how to lock that down.
- Notes on Gartner's 2018 Data Science and and Machine Learning MQ (ML/DL): Not only is this an insightful take on this Magic Quadrant, but it's also entertaining. Well, as entertaining as a post breaking down a Gartner report on data science can be ;-)