We already have news about the arrival of the expected and Nueva Siri. They are not good. Apple points to the spring of 2026 as a launch objective for its new assistant, with the aim of bringing it as part of the update to iOS 26.4, according to Bloomberg sources.
The delay is completely in line with the recent interview of Wall Street Journal to Craig Federight and Greg Joswiak: Apple knows that he is losing the AI career, and he only has to trust that his proposal ends up working in the long term.
They painted you little birds in the air. It has been just one year since that presentation of Apple centered on Apple Intelligence. One in which they told us about an integration of Siri in the system as we had not seen: an assistant with the ability to understand each corner of iOS, with an advanced contextual understanding and integration with each native app of the system.
In principle, the final deployment was planned for autumn of 2024, but the delays accumulated and accumulate until they reached a practically unsustainable point.
iOS 26 as first real step. Apple took advantage of the WWDC 25 to present iOS 26, one of the biggest changes at the visual identity level ever seen in Apple’s operating systems, now much more unified. Next to him, artificial intelligence functions applied to everyday use arrived: although for practical purposes they do not go far beyond the translation of texts, filters for calls and some improvements in the generation of genumjis in Image Playground.
After almost a year of delays, Apple finally implied what is its philosophy with AI: work to make it local, but with a functioning similar to that of its rivals.
More delays. After the presentation of iOS 26, with novelties in AI but not a single word about the new Siri, the question was forced. What is happening with her? The responses of a Craig Federighi, nothing height but not entirely comfortable, perfectly revealed the moment in which Apple is.
- (Joanna Stern) “Siri is not better than his competition”, (Craig Federight) “already, but it will be, it is our mission.”
- (Greg Joswiak) “It would be disappointing to launch something that does not meet our quality standards.”
- (Craig Federight) “This is a new technology, nobody is doing very well right now.”
Joswiak, software leader in Apple, justified its road map around the key point: the company wants its AI to be discreet and that the user can perform tasks with the phone without even realizing that he is using Apple Intelligence. In fact, they do not want a dedicated app, as there is for Gemini or Chatgpt. The problem? Google has achieved this goal for a long time.
Apple’s rivals are doing well. Although Apple points to the immaturity of the AI for mobiles, the truth is that one of its best moments lives. And phones such as the Google Pixel or the Samsung Galaxy S25 Ultra are the best proof of this. Here you have to make a distinction between Gemini as an app integrated in the system, and with Gemini Nano as a language model. On a phone like the Ultra S25 you can:
- Translate real -time calls without knowing that AI is doing.
- Remove the background noise from a video automatically, without Gemini telling you what you are doing.
- You can transcribe a voice recording with a single click, without opening additional applications.
- If they call you, the phone will automatically detect whether or not spam, without notices about ia.
Yes, Google (although its approach is not so local), is able to integrate Gemini silently into the system. So much so that their functions are still native to the system, and the user does not have to know what is done with Ia and what not.
Apple needs to overcome Gemini at AI. It will not be easy.
Apple, for now, has not shown to live. During Apple’s presentation, its action fell 1.5%. The expectations were high, but the presentation was a clear Apple message showing not being above any of its direct rivals, and relegating third parties (OpenAi) the most advanced functions of AI.
With the looks put in 2026, the pressure to which Apple is subjected is even greater. The maturity point of its main rival, Google Gemini, is very high, above proposals in Benchmarks with respect to Grok 3, O3-mini, Deepseek R1, Claude 3 and calls 4. To recover Apple confidence needs results, not promises.
Image | WorldOfSoftware
In WorldOfSoftware | We have discovered something worrying in the AI models: if the problem is too difficult, they give away immediately
