Every top tech company has announced its interest in conversation as a way to interact with computers this year. Microsoft announced its Bot Framework, Facebook opened Messenger to developers, Google announced Allo, and Amazon keeps expanding Alexa. Until recently, one tech giant was missing: Apple.

At their WWDC conference, Apple finally announced its entry into the conversational interface space. But the company did it a perplexing way. Both its Siri and Messages announcements offered a hint of open integration, but in unexpected ways that were not comprehensive. The primary reason for the limitations was not user experience considerations, the usual justification for its similar restrictions. Instead, Apple faces serious strategic threats that prevent them from fully embracing conversational interfaces.

Siri, extended. Sort of.

Apple’s Siri announcements were simultaneously exciting, disappointing, and stunningly predictable.

Developers get a way to integrate with Siri. This is something third-party developers have requested for years. The potential for innovative services via Siri is exciting, but details matter.

The nature of the Siri integration is underwhelming. Apple’s SiriKit  —  what they call their tools for integrating with Siri  —  is limited to specific use cases, and it restricts how a developer can integrate Siri into their own products.

Siri was opened in a way that reinforces the practice of installing an app on your Apple device. As a consumer, to add functionality or features to Siri, you must install an iOS app. As a developer, to integrate with Siri, you must create a special addition to your iOS application that enables Siri to tell the iOS app when the user wants to do something. Siri extensions cannot be built or added outside the context of an iOS app.

SiriKit does not allow a developer to leverage Siri’s language processing or voice recognition capabilities for arbitrary purposes. Developers may create an iOS app that fulfills the wishes that Siri determines the user has. But they are not allowed to do so outside a few use cases, and even within those use cases they are restricted in how innovative they can be, since Apple has predetermined what the conversations can look like.

Some common domains are clearly ripe for third-party Siri integration, like mapping and music. Yet these have been excluded from third-party innovation. The most likely explanation seems to be that integrations with services like Google Maps and Spotify might compete with Apple’s existing solutions like Apple Maps and iMusic.

What an open Siri would look like

A completely open Siri would look quite different from what was announced. While downloaded apps might still play a role, the capabilities offered to developers would go far beyond a limited set of domains — they would give developers the power and freedom to be truly innovative in every domain.

Indeed, an open Siri might look very similar to Amazon’s Alexa. Alexa’s third-party skills are not limited in the type of service they can implement. Developers are free to be innovative in any domain. Despite that freedom, users access the extensions to Alexa’s core in a relatively seamless way. The end user only needs to reference the name of the third-party service in his or her interaction.

Overall, Alexa is a very different service from Siri, however. Alexa is extended through connections between Amazon’s servers and those of the third-party developers, not installed apps. That is because Alexa is a hosted cloud service and is not tied to a specific device.

It is unlikely that Apple would simply offer the language processing technology behind Siri to developers. That would remove too much control over the end user experience from Apple.

Apple Messages

The Apple Messages announcement was similarly predictable in its implementation strategy. Like Siri, third-party enhancements to Messages require users to install conventional iOS apps. These apps will enhance the functionality of the Messages app running on their phone. For example, as you are typing to another person in Messages, you might get access to a relevant app embedded within the Messages app. This is convenient, but definitely not the same type of conversational interaction that Microsoft, Facebook, and Google are headed towards.

If Apple were to embrace conversational interaction the way other large companies have, it would open iMessage itself to developers. There was speculation that Apple might do that, and even extend it beyond just Apple devices. Neither happened.

The Messages integration Apple announced, and its restrictions on iMessage, reinforce the need to own an Apple device. Again, strategic and competitive positioning forced this outcome.

Apple does have reason to open iMessage to developers. Facebook opened up Messenger because allowing companies to offer services within Messenger meant that users would spend more time in Facebook’s products. If users are able to shop, get information, and be entertained within Messenger, they will cleave to Messenger more and alternatives that do not allow these activities, like iMessage, less. Apple could have taken the same approach with iMessage to stay competitive here. But it did not.

Why iMessage isn’t open

Apple needs to maintain its hardware sales, and that means maintaining differentiation between the hardware it sells and that of its rivals. The reason people buy an Apple phone instead of the hundreds of other products is the unique benefits that Apple’s iOS and its ecosystem offer: user interface, consistency, and simplicity.

Enabling users to do more activities via messaging platforms, even Apple’s own iMessage, could erode the competitive advantage of iOS and Apple’s devices. Conversational services — whether delivered via Facebook, Google, Telegram, Kik, or iMessage — make downloading mobile applications less necessary.

Conversational apps are poised to replace many native mobile apps. They will transform common activities, from shopping to money management, even if they might never be suitable for such activities as games and photos. And no app download is required. For those activities, users will not even have to leave messaging apps, meaning that the operating system surrounding the messaging interface becomes less and less important.

A reduction in the use of installed mobile apps due to the evolution of messaging would seriously threaten Apple’s primary business by eroding the differentiation that iOS offers, threatening their hardware sales. Apple therefore cannot promote the displacement of mobile apps by conversational services yet.

Apple’s difficult position

If conversational interfaces reach the potential that Apple’s competitors envision, Apple will face difficult choices. Given the current landscape, Apple could need to either broaden its revenue streams beyond hardware or figure out a way to compete here with Google, Microsoft, and Amazon.

Today, Apple is not well positioned for the conversational interaction battlefield. Conversational services tend to be delivered via hosted cloud services. And as they mature, conversational services will become more and more reliant on advanced machine learning technology as user experience expectations move beyond simple bots towards full natural language. Apple’s competitors are stronger here.

Apple’s ventures into cloud applications have generally been “me too” — defensive offerings like iCloud and Apple Music. They are not Apple’s strength, which is understandable because cloud services have not been a core revenue stream for the company.

In the field of machine learning, Google, Microsoft, and Facebook all have substantial natural language processing research and technology endeavors. That is not to say Apple cannot make advances in machine learning if it becomes a priority: Amazon started at a relative disadvantage as well and was able to acquire high-caliber researchers and achieve excellent results on speech recognition with their Alexa services.

Pure research capability is not necessarily a requirement for delivering great conversational products. Apple certainly has the ability to take existing research and apply it at scale to deliver excellent products. In fact, it is startups with limited resources but tight product focus, like Api.ai, Init.ai (of which I am a founder), and the once-independent Wit.ai, that are advancing the frontier of conversational user experience today, not the larger companies with more substantial research budgets. But Apple does not have a competitive advantage in conversational product design right now.

The broader issue is not that Apple is behind in the fields of cloud services and machine learning: It is that Apple does not have an inherent advantage in either. In the most optimistic light, the company is only on par with its competitors. Entering a market without a clear way to outcompete would be uncharted and precarious territory for Apple. So, for now, Apple needs to prevent conversational interaction from developing into a battleground on which it is ill-equipped to compete. That is why Siri, Messages, and iMessage are so limited.

The unresolved question is where Apple goes from here. Its competitors are unencumbered by Apple’s business model considerations and have a head start in the relevant expertise. Apple seems to be publicly marketing that its differentiator is its respect for privacy and protection. But in a world where privacy seems to matter less and less to consumers, only time will tell whether this strategy will be sufficient.