Siri has spent years training iPhone owners to stop asking it for help.
That is a strange legacy for one of the best-known assistants in consumer tech. Ask the old Siri to send a text, set an alarm, or add a reminder, and it might work. Or it might misunderstand you, stall halfway through, or decide the web was somehow the answer.
After living with the iOS 27 beta since day one, I’m convinced Apple has finally addressed that basic problem.
This is not a review of Apple Intelligence in its entirety. It is an early review of Siri AI: the new assistant Apple is building into iOS 27.
ChatGPT Voice still has the edge in intelligence, open-ended conversation, and the feeling that you can hand it a messy question and genuinely think out loud together.
But Siri is finally becoming useful as a phone assistant. That is a much bigger deal than it sounds.
- The small requests are where Siri has changed most
- Siri AI is not ChatGPT Voice and that is fine
- More natural to speak to, with a few important caveats
- Apple Watch is the quiet beneficiary
- The big promises I am not ready to judge yet
- Third-party apps remain the obvious weak point
- The privacy argument deserves a mention
- Early verdict: Siri is becoming worth using again
- Related reading
The small requests are where Siri has changed most
The new Siri is faster to recognize a command, quicker to respond, and much less likely to turn a straightforward request into a tiny administrative battle. Messages, email drafts, reminders, alarms, and other native Apple tasks are the things I now actually ask it to do instead of just doing them myself.
That willingness is the real test.
The old Siri could technically handle plenty of these tasks. It just handled them inconsistently enough that people learned not to rely on it. iOS 27 is beginning to reverse that habit.
One night, I asked it to turn on every alarm I had set between 5 a.m. and 6 a.m. It did exactly that. Later, I could holler at it to turn them back off. Was that the best call for my morning motivation? Debatable. But it was precisely what I asked for, which is a low bar Siri spent years finding inventive ways to limbo under.
This is not a flashy benchmark result. It is better: it is the kind of quiet reliability that makes an assistant part of a routine. Apple didn't need Siri to compose a sonnet about my sleep schedule. It needed Siri to manage the alarms I already made.
Siri AI is not ChatGPT Voice and that is fine
ChatGPT Voice remains my preferred voice assistant for intelligence and conversation. It is better at working through a topic over several turns, exploring an idea from different angles, and responding to questions with no clean, transactional answer.
Siri’s job is different. Siri has the keys to the house.
It can work inside the phone’s native apps and settings, where a standalone chatbot cannot easily go. Apple’s big iOS 27 bet is that the assistant closest to your messages, mail, photos, calendar, reminders, and device controls can win by being more useful—not necessarily by being the best conversationalist on every topic.
That argument is starting to hold up. If I want to brainstorm, research, or reason through something, I still reach for ChatGPT. If I want my phone to do something, I am increasingly comfortable asking Siri.
There is no shame in that division of labor. In fact, it is probably the more honest vision of what a practical assistant should be.
More natural to speak to, with a few important caveats
Siri also feels more conversational. It is gentler in tone, more forgiving of imperfect phrasing, and less mechanical to use while walking, driving, or making coffee.
Apple offers a healthy selection of voices. I settled on Irish 2. They are not the best synthetic voices I have heard, but they are easily good enough to make the experience feel more human and more pleasant.
Apple is also promising adjustable voice pace and expressiveness, alongside improved systemwide dictation. Those upgrades fit what I am seeing: the interaction is not merely faster; it has less friction. But they are worth treating with a hardware asterisk. Apple says its most advanced on-device model—and some of the most expressive voice and dictation capabilities—require its newest, higher-memory devices. Not every Apple Intelligence-capable iPhone gets an identical Siri experience. Apple’s launch details are unusually specific about that split.
Apple Watch is the quiet beneficiary
The experience is especially good on the Apple Watch. I am also running watchOS 27, and the tighter integration makes sense. A watch is often the place where voice control has the least competition from tapping: you are carrying something, moving, exercising, or simply trying not to pull your phone from a pocket.
When Siri responds quickly, the watch becomes a useful command surface for brief asks. When Siri is unreliable, it is just a tiny screen that makes you repeat yourself in public. So far, the iOS 27 and watchOS 27 combination is landing much closer to the former.
The big promises I am not ready to judge yet
This is where the beta framing matters.
Apple’s headline Siri AI features go well beyond alarms and reminders. The company says Siri can use personal context to find a restaurant recommendation in Messages, surface a confirmation number from Mail, understand what is on your screen, and take action based on that context. It also has a dedicated Siri app with conversation history that privately syncs across devices, plus a new Camera mode for Visual Intelligence. Apple’s WWDC overview makes clear that these are the center of its pitch.
Those are also the capabilities I would need to test more deeply before declaring Siri transformed.
They are the real measure of whether Apple has built an assistant or merely improved voice control. Finding a hotel confirmation from an old email is compelling. Taking the next sensible action after finding it is more compelling still. But that is a much harder problem than turning on alarms, and early beta software is exactly where the gap between a good demo and a dependable feature tends to show up.
For now, I would frame the strongest claim this way: the foundational layer is working. Siri feels faster, more reliable, and more natural for the native tasks people already expect it to handle. The genuinely agentic layer—personal context, screen awareness, and broader app actions—needs more time and more testing.
Third-party apps remain the obvious weak point
The limits are still clear when a request becomes complicated, requires several conditional steps, or reaches outside Apple’s ecosystem.
Ask Siri to build a Shortcut, orchestrate a more elaborate workflow, or take action inside a third-party app that has not been deeply integrated with Apple’s system, and the old limitations start to show through.
That is not only a model-quality problem. It is an ecosystem problem. Siri’s reach into third-party apps depends on developers exposing useful actions and making their content legible to Apple’s system. As The Neuron’s guide to Apple Intelligence in apps explains, developers have to do that work before Siri can understand an app well enough to act inside it.
So, for now, Siri is best at familiar Apple-shaped tasks and less reliable in open territory. It is a better executive assistant for your iPhone than it is an all-purpose AI agent.
The privacy argument deserves a mention
This is also where Apple has a different proposition from ChatGPT Voice. Apple says Siri AI runs work on-device where possible and uses Private Cloud Compute for more demanding requests, with personal data not stored or accessible to Apple when that cloud system is involved. The assistant’s new conversation history can sync through iCloud across devices. Apple’s architecture explanation is part of the company’s case for why Siri should be trusted with more personal context.
That does not mean users should stop thinking critically about what they share with an assistant. It does mean Apple is competing on something more substantive than a new voice or a flashier animation: it wants the assistant with the most intimate access to your digital life to also feel like the safest place to use it.
Early verdict: Siri is becoming worth using again
Apple has meaningful beta time left before iOS 27 reaches the public. The feature set, regional availability, and supported hardware are all still moving targets. Apple Intelligence and Siri AI require an iPhone 16 or later, or an iPhone 15 Pro or Pro Max; Siri AI is also initially unavailable on iPhone and iPad in the EU, and Apple says the broader feature set will not be available in China while it works through regulatory requirements. Apple’s availability notes matter because this will not be a universal iPhone upgrade.
But the early verdict is straightforward.
Siri AI has not displaced ChatGPT Voice as my preferred assistant for intelligence or conversation. When I want to explore, brainstorm, or reason through something, ChatGPT still gets the call.
When I want my phone to do a thing, though, Siri is finally starting to earn the ask.
For Apple, that is a much bigger win than it sounds. The company has spent years selling Siri as a convenient helper while users quietly learned to stop relying on it. In the iOS 27 beta, Siri is starting to make the case that it belongs in the routine again.
Related reading
- Everything (AI) Apple announced at WWDC 2026 — a broader look at Apple’s system-level Apple Intelligence strategy.
- Apple Intelligence in apps: WWDC26 developer guide — why third-party support is the next major test for Siri AI.
- Honest review of ChatGPT Voice Mode — useful context for the conversational standard Siri is now chasing.