Apple operating systems focus on optimization and AI for 2027

As expected, Apple unveiled the next versions of the operating systems powering its computers, smartphones, tablets, media players and smartwatches during the WWDC 2026 keynote – and, just as expected, it largely focused on three things: AI, voice assistance and code optimization. That last one has been in many consumers’ wish lists for a long time, as macOS, iOS and iPadOS needed performance improvements, extensive bug fixing and fine-tuning way more than new functionality or features. Apple finally listened, promising increased speed in all kinds of everyday operations – not just on newer devices, but on older ones too, all the way back to the iPhone 11, the iPad Pro 4th-gen or the M1 Mac models.

We’ll have to wait until the end of September to get a clear picture of how the final versions of macOS 27, iOS 27 and iPadOS 27 actually perform in everyday use. The very fact, though, that Apple openly stated it worked based on consumer feedback for requested changes, additions or fixes is encouraging.

Speaking of fixes, it seems the company has now accepted that (a) not everyone liked the Liquid Glass user interface introduced last year and that (b) people simply can’t agree on what’s the “correct” amount of transparency, reflection or refraction effects for iPhone or iPad screen… so it’s officially leaving it to consumers to decide by introducing a slider capable of adjusting these effects in small increments.

To be fair, Apple – based on the keynote videos and the developer presentations that followed – seems to have done a lot of boring but necessary work addressing UI inconsistencies, fixing small but irritating bugs and adding various quality-of-life improvements in many areas of its operating systems.

This is exactly what yours truly thought that these versions of iOS, iPadOS and macOS needed, so it’s great that Apple took the time to do what its actual customers have been asking for, instead of just doing what its shareholders wanted (focus on AI for entirely different reasons). Not that the latter did not get their wish, of course, but at least we got ours too. That’s… something, no?

Apple Intelligence v2.0 comes in various forms, not everyone gets all of them

Most consumers may be happy with the performance and quality-of-life improvements of the next Apple operating systems but in the context of this particular WWDC many of us were eager to find out how the company made use of Google’s Gemini models in this new Apple Intelligence implementation. In a word: it did so well, even if there are some limitations leading to unforeseen complications people need to be aware of.

Apple Intelligence 2.0 is basically playing catch-up to what companies like Google, Samsung and others have been experimenting with on Android for the past three years or so: smarter writing tools, contextual messages and notifications, visual recognition, intelligent and creative photo tools, as well as a more capable Image Playground function, they are all here and work pretty much as one expects them to. We’ve seen it all before, yes, but Apple has gone the extra mile in terms of how this functionality blends into iOS, iPadOS and macOS: it is done in a natural, effective way that most people will not object to.

This is key, even if it does not guarantee in any way that consumers will actually use all of these AI tools in everyday life. Many of them, unfortunately, may not even have the option: as it was widely suspected, Apple Intelligence v2.0 is based on multiple AI models – no less than five different ones, in fact. Two of them run on-device and the rest on either Apple’s or Google’s servers. The base on-device AI model runs on most Apple smartphones, computers and tablets quickly and efficiently, while the advanced one needs 12GB of RAM at minimum (so it only runs on the latest and most capable iPhones and iPads).

This may, in time, prove to be a problem: the “Apple Intelligence” term currently applies to all AI functionality, but the available range of those AI functions themselves – as well as how fast they are in operation – will vary from model to model and from device to device going forward. This can easily lead to consumer confusion and it will be admittedly difficult for Apple to explain (especially to non-technical customers) how or why some iPhones or iPads can do certain AI things while others can’t.

In some respects, this is the price of evolving hardware all manufacturers have to pay at some point. In others, it could surely disappoint owners of e.g. an iPhone 16 Pro Max or an M4 iPad Pro with 8GB of RAM – yours truly owns both – to find out that they won’t be able to run the most advanced version of Apple Intelligence come September. It will be, in any case, interesting to see how Apple handles this complicated matter long-term.

Siri AI has a lot to prove going forward but there’s reason to be hopeful

The long-awaited reboot of Apple’s voice assistant, Siri, was the other thing many of us were curious to see – especially after the 2024 promises the company made but never kept – and, this time around, the shining orb did not disappoint. It now “lives” within the Dynamic Island on iPhones (so no wondering what Apple will do with it anymore) but it can be called in all the usual ways, as well as through Spotlight on Macs and even just by looking at it in Vision Pro. It is, in other words, as deeply integrated to Apple’s operating systems as ever, only it can now work as both a voice assistant and a chatbot, always powered by AI.

Because it is based on the LLM models Apple developed in collaboration with Google, Siri AI can now do much more than just answer questions: it can provide all kinds of information in conversational mode, it can interact with apps, it can suggest contextual actions and more. It’s worth noting that there’s also an official, stand-alone Siri AI app coming, where consumers will be able to find past conversations with Apple’s revamped assistant, complete with iCloud syncing for access from other devices at a later time.

Noting that Siri AI has a lot to prove would be an understatement: this is probably the virtual assistant that Steve Jobs envisioned more than 15 years ago (when he acquired the Nuance-powered Siri back in 2010). This is the one major iPhone function Apple failed to deliver on for over a decade, despite millions of dollars spent in software and services development, so the company owes it to its customers – as well as to itself – to make it work.

It always felt like Siri was missing a few essential building blocks in order to work as intended and maybe Google’s AI models – along with fast network access – was exactly what it needed. Will it meet consumer expectations? That’s a question people will be able to answer for themselves in the fall – unless they live in Europe or China, unless they speak any language other than English or unless they don’t use beta software: Siri AI won’t be available in those markets and will probably not be officially complete in 2026. Funny how this works with almost any completely new or advanced software feature Apple brings to the table, no?