The Siri Shortcut

Shortcuts is an overloaded term in the Apple world. What I am referring to in this article is the Siri Shortcuts system, not the Shortcuts app that allows users to make their own workflows. You can make Siri Shortcuts from shortcuts made in the Shortcuts app.

What I am focusing on is the mechanism by which apps, first or third-party, can expose shortcut actions to iOS that can then be ‘added into Siri’ and activated using your voice. The shortcuts can be presented to the user by each individual app or found in a list of recommendations in the Siri pane of the Settings app.

This means Siri can now be smarter by drawing on the capabilities of many more apps. You can order coffee. Control third party audio apps like Overcast or Pandora. Plan travel itineraries with Kayak. All with your voice talking to your intelligent personal assistant.

Except that’s not really true. That is how Apple likes to market the feature but it’s a twisted form of reality. Shortcuts are not making Siri smarter, in fact they are dumber than pretty much anything Siri has done to date. Shortcuts put the burden on the user to do the legwork of synthesising data sources and integrating the apps into the voice service.

Shortcuts require registration and administration to do anything at all with Siri. The user has to pre-emptively search out every command available in a certain app and then add each in turn to Siri. Registration requires the user to think up the phrase they want to use to trigger the command on the spot. Siri can then trigger these actions when that same phrase is said back to it at a later date.

There is no intelligence here. Siri transcribes the user’s voice and looks for an exact text match of that phrase in the database of voice shortcut phrases that the user has generated off their own back. If a match is found, it proceeds. Otherwise, failure.

There is no leeway in what can be asked. There is no flexibility in how a command can be impromptu formulated. Any sequence of words other than what was inputted when the shortcut was first registered is not understood. That’s the point, there is no understanding. The Siri interpreter has no understanding of semantics or meaning when you are interacting with an app shortcut. It is a dictionary text lookup and nothing more.

It is disappointing that Apple is leaning so heavily on shortcuts as a mainstream way for customers to get more from Siri. It flies in the face of how you want a voice assistant to work and behaves differently from every other type of Siri interaction. When you ask Siri for the weather, you can say ‘What’s the weather?’ or ‘What’s the weather on Friday?’ or ‘tell me the forecast’ or ‘do I need to wear sunglasses?’ or just ‘weather’. The whole point is the user does not have to revise a set list of triggers. Apple has made entire ad campaigns to this effect, promoting the flexibility. Forget custom variables, the Shortcuts system cannot support multiple ways of saying the same thing. A truly good voice assistant does not require the user to remember something.

This philosophy is exactly what drove Apple to design the SiriKit API in the way they did. SiriKit abstracts aways the parsing or semantics of a snippet of speech. It is the responsibility of Apple engineering to enable the understanding, across English locales and foreign languages. Under SiriKit, the third-party apps only supply the data for the response. All of the work Apple puts in to improve Siri’s understanding of commands automatically benefits every SiriKit app, and every SiriKit app of the same domain should respond consistently to the same commands. The downside to SiriKit is that it can only work with a subset of applications, those which Apple has done the legwork to create a domain for. So, a user has to know which of the apps on their phone works with Siri, but they don’t have to register a corpus of commands and can interact with the app through Siri with an order of magnitude more freedom.

By relying on the Shortcuts system so heavily, Apple risks breeding even more hatred for Siri than what already exists in the community, as people inevitably forget the exact words they used when they made the shortcut four months ago, and blame it on Siri being dumb instead. If you want to call Siri dumb, then Shortcuts is primordial in comparison.

I am worried that Siri Shortcuts has usurped favour in Apple’s product marketing groups to the point where engineering resources on SiriKit have been sidelined. Siri Shortcuts certainly requires less effort on the part of Apple to maintain. It’s the lazy way out for them versus SiriKit which necessitates continuous advancement and development of new domains. I care about the end user experience and SiriKit provides superior results. I want them to invest in supporting more domains. At best, Apple should consider Siri Shortcuts as a companion feature to Siri and not a substitute for SiriKit.