Phone assistants that do things for you: the 2026 shift

From answering to doing

For years, phone assistants could set a timer, play a song or answer a simple question. The big change in 2026 is that they are starting to do things on your behalf: read what is on your screen, pull details from your messages and emails, and carry out a chain of steps across different apps.

The technology behind this is the same kind of AI that powers chatbots, called a large language model: a program trained on huge amounts of text so it can understand requests written or spoken in everyday language. When an AI is allowed to take actions, not just chat, people often call it an agent, a bit like giving a helper the keys to run errands for you.

What Apple launched

Apple announced a rebuilt Siri, which it calls Siri AI, at its developer conference in June 2026. Reports at the time said it is built on Google's Gemini models. It arrived with iOS 27 and Apple's other September software updates, as a beta (an early, still-being-tested version) in English, with French, Japanese, Korean, Portuguese and Spanish due in October.

According to Apple, the new Siri can:

• Use personal context from your messages, emails and photos to answer questions.
• Answer questions about whatever is on your screen.
• Search the web for current information, and keep a conversation history in its own Siri app.

It needs a fairly recent device, such as an iPhone 15 Pro, an iPhone 16 or later, or an M1 or newer iPad or Mac. Apple said it is not available in China, and in the EU it was initially limited to Mac and Vision Pro. Apple also said features that run on its servers have daily usage limits, and that expanded access will cost money in future.

What Google is doing on Android

In May 2026 Google announced new Gemini features for Android that can handle multi-step jobs. Its example was copying a shopping list from a notes app and adding the items to a shopping app's basket. You start a task by holding the power button and describing it, and the assistant uses what is on your screen as context.

Two design choices are worth knowing. Google said Gemini will wait for your final confirmation before completing a checkout. And filling in forms using details it has learned about you is opt-in, and can be switched off in settings. Google said the features would reach the latest Samsung Galaxy and Pixel phones first, then other Android phones later in the year.

What it means for ordinary people

The good: these assistants can save real time on fiddly jobs, such as finding the address someone texted you, turning a recipe into a shopping list, or summarising a long email thread. Speaking a request can help people with limited mobility or eyesight, or anyone who finds small screens hard.

The cautions:

• They still make mistakes. AI can misread a request, pick the wrong contact or invent a detail. Reviewers have described the new Siri as a beta for a reason.
• They see a lot of personal data. To be helpful they need access to messages, emails and photos. Read the privacy settings and decide what you are comfortable with.
• They can be fooled by what they read. An assistant that reads web pages or messages could be misled by text planted there by a scammer. Treat anything it suggests doing with money or passwords with care.

Sensible habits

1. Start small. Try low-risk tasks first, such as reminders, summaries or finding a photo.
2. Check before you confirm. Read the recipient, amount and details before anything is sent, booked or paid.
3. Review permissions. In settings, see which apps and data the assistant can use, and turn off what you do not need.
4. Keep your phone updated and locked. An assistant that can act for you is another reason to use a strong screen lock.

Want a gentler introduction to the technology itself? Our Learn section explains how chatbots work and how to use AI tools safely. For the bigger picture on trust, see what AI means for online trust.