AI on our phones is getting a lot more interesting! What started with photo tricks, chatbots and the odd novelty feature is now starting to become part of how we actually use our phones day to day.

Newer tools can understand what’s on your screen, work out what your camera is pointing at, pull useful information from messages and emails, translate calls, remember things for you and even help with tasks across different apps. It’s moving quickly — and some of the most useful features are starting to feel less like “using AI” and more like your phone simply being a bit smarter.

Not everything is available on every device yet, and some features still depend on the phone, region or software version. But there’s already a lot more happening than you might realise.

Your Phone Can Now Understand What’s on Your Screen

One of the biggest changes is that AI can increasingly understand the thing you’re already looking at instead of making you copy, paste or explain it first. Apple’s new Siri AI can answer questions about what’s on screen and take actions from it. Apple gives the example of looking at a sports website, asking Siri to find the upcoming home games, then adding them straight to your calendar. It can also use personal context from messages, emails and photos to help find information you already have somewhere on the phone. Siri AI started rolling out in beta in English in September 2026. 

Gemini works in a similar way on Android. Gemini Live can see a shared screen, so instead of trying to describe the confusing bit of an app or a page you’re stuck on, you can show it and ask for help directly. That might sound like a small change, but it removes a lot of the friction that made phone assistants feel more trouble than they were worth.

The Camera Is Starting to Become an AI Input Too

Your camera isn’t just for taking photos anymore. With Gemini Live, you can point the camera at something and ask questions about what you’re seeing. Google suggests uses ranging from understanding a blinking error code to getting help with instructions or organising something in front of you. Its newer Guided Vision feature can also describe surroundings and help users frame objects correctly, although Google makes clear that it isn’t intended as a navigation or mobility aid. 

Apple is doing something similar with Siri mode in the iPhone camera. Siri AI can interpret what the camera is seeing and help take action from it, rather than making you photograph something and then go off to search for it separately.

This is probably one of the more genuinely useful directions for phone AI because it cuts out the awkward bit where you have to explain an object, message or problem in words before the assistant can help.

It Can Find Information Buried Across Your Phone

Anyone who’s ever searched messages for “that restaurant Lindsay sent me three months ago” will understand why this matters! AI assistants are increasingly able to pull together information from different parts of your phone. Apple says Siri AI can use personal context across Messages, Mail, Photos and other apps. In one of Apple’s examples, Siri can find an activity mentioned in a message, locate a related recipe from an email and then add the ingredients to a Reminders list. 

Google’s Magic Cue on Pixel 10 takes a similar approach. With permission, it can surface useful information from Gmail, Messages, Calendar and other sources at the point you need it. If you’re calling an airline, for example, it can bring up the relevant flight information instead of making you hunt through your inbox while you’re on the phone. 

This is where AI on phones starts becoming more interesting than simply asking a chatbot a question. The useful bit is that it can potentially understand the context of what you’re doing and find the information without you remembering where you stored it.

It Can Translate Calls While You’re Having Them

Real-time translation is another area that has become much more practical. Pixel’s Voice Translate can translate phone conversations in real time and recreate the translated speech in something resembling each person’s voice. It currently supports a range of languages when translating to or from English, including French, German, Japanese, Spanish, Portuguese and Indonesian. That’s a very different experience from copying a sentence into a translation app, waiting for the answer and then trying to continue the conversation.

Availability still varies by device and language, but this is one of those AI features that has a fairly obvious real-world use rather than needing much imagination to justify it.

Your Phone Can Take Notes From Calls for You

Another useful development is AI handling the information that comes out of a phone call. On supported Pixel phones, Call Notes can record a conversation, create a transcript and summary, and suggest follow-up actions such as calendar events, tasks or reminders. Google has also expanded the feature into more markets over time. That could be useful after calls about appointments, deliveries, travel changes or anything else where you’d normally finish the conversation and immediately realise you’ve forgotten half of what was said.

There are obvious privacy considerations here, and recording rules can vary depending on where you are, so this isn’t something to switch on thoughtlessly. But used appropriately, it’s a good example of AI taking a slightly annoying admin job off your hands.

It Can Remember Where You Put Things

This one is wonderfully ordinary. Google’s September 2026 Android update added the ability to ask Gemini to remember where you’ve put something, such as a passport or spare key. It can save that information into Find Hub, including an optional photo, so you can ask for it later instead of relying on the traditional system of “I definitely put it somewhere sensible”. That’s arguably more useful than many of the flashy AI demos because it solves a problem people actually have.

It also shows where phone AI is heading: less about generating something impressive and more about quietly remembering, finding and organising information for you.

It’s Starting to Do Things Across Apps, Not Just Answer Questions

The next step is AI actually carrying out multi-step tasks. Google has been previewing Gemini features that can handle selected actions across apps, such as helping book a ride or reorder a previous meal. These features are still limited, with compatibility varying by device, country and app, and Google explicitly describes some of them as beta experiences. Apple is moving in the same direction. Siri AI now has wider system-level actions and can do things like draft an email or edit and share photos, with broader third-party app support also planned. 

This is probably the bit worth watching most closely. There’s a big difference between AI telling you how to do something and AI actually doing several of the steps for you.

It Can Help You Take Better Photos, Not Just Edit Them Afterwards

AI photography is also becoming less about dramatic edits and more about helping before you press the shutter. On Pixel 10, Camera Coach uses Gemini models to suggest framing and composition changes while you’re taking a photo. Gemini Live can also provide visual guidance on screen while the camera is open. That feels more useful than another filter because it’s helping with the photo you’re actually trying to take rather than trying to rescue it afterwards.

Of course, whether anyone wants their phone critiquing their composition during dinner is another question.

The Useful Stuff Is Getting Less Flashy

The most interesting thing about phone AI right now isn’t necessarily image generation or having long conversations with an assistant. It’s the smaller jobs. Finding the flight details buried in your email. Remembering where you put your passport. Translating a phone call. Understanding what’s on your screen. Looking at something through the camera instead of trying to describe it. Pulling the useful bits out of a call so you don’t have to scribble them down.

That’s probably where AI on phones starts making the most sense: not as another app you remember to open, but as something that quietly removes a few annoying steps from the things you were already doing.