Siri AI Opens to Every App: What Marketers Need to Know

Overview

Apple has opened Siri AI to every app. On November 3, any developer gained a way to plug an app's content and actions into Siri AI through a new Developer Preview in the iOS 27.3 beta, Apple said in its developer news.

That sounds like a developer story. It's also a marketing one. If people can ask Siri to check an order, book a table or buy something again, they may not open your app, see your home screen or scroll your offers first.

There's a catch, though. The preview only runs in test builds, and App Store versions of apps can't use it until submissions open in the spring.

So you have a few months to prepare. This guide covers what Apple confirmed, what's still unknown and what marketing teams should do now. Disclosure: Vanaxity uses AI models in our own work, and this article was drafted with help from Claude. We cite our sources and label our own estimates.

Key takeaways

• Since November 3, 2026, any developer can connect an app to Siri AI in the 27.3 betas of iOS, iPadOS, macOS and visionOS.
• Apps can expose their own actions and content, not only Apple's preset templates. Examples include checking a delivery, booking a table and placing an order.
• No new entitlement or special program is needed, but it only works in development builds and internal TestFlight for now.
• Customers get the first features in a few months. App Store submissions open in the spring.
• There's no Siri ad product, no ranking data and no published terms yet. Plan tests, not budgets.

Siri AI developer preview at a glance

What changed on November 3

Siri AI isn't new. Apple showed it at WWDC on June 8, and it shipped in beta in English with iOS 27 on September 14. Back then, only a few outside apps worked with it. WhatsApp and Audible were the first.

What's new is the scope. Until now, Siri AI could only act inside other apps through app schemas, which are fixed templates Apple wrote for common jobs like playing audio or sending a message, PPC Land explained.

If your app did something outside those templates, Siri couldn't reach it. Now it can. Any app can describe its own actions, and Siri AI can call them when a request fits.

Apple also lets Siri AI find anything an app adds to the Spotlight semantic index, the on-device index of app content, Unite.AI reported. That includes items that don't match any preset template.

From 91 preset actions to any action

How big is the jump? Apple didn't give a number for the old limit. FourWeekMBA counted 91 preset actions across 13 main domains in Apple's documentation, plus 85 more in 9 domains that only worked in Shortcuts.

Treat those numbers as one outlet's tally, not an official figure. The point holds either way. The old list was finite, and it was written by Apple.

The new model is open-ended. A pet store can expose "reorder my dog food." A gym can expose "book a spin class." Neither job fit a template before.

How apps plug into Siri AI

Apps connect through App Intents, the framework Apple already uses for Shortcuts and widgets. A new Swift macro, @AppIntent(.tool), turns a custom action into a tool Siri AI can call, per Unite.AI's reading of Apple's docs.

Developers stay in control. They choose which content Siri may read and which actions it may take, and they can lock sensitive fields behind Face ID or a passcode or hide them from Siri entirely.

• Actions: book, reorder or cancel.
• Content: items Siri can find and show, like orders, bookings or saved products.
• Snippets: small cards that show the result in your own design and branding, so an answer still looks like your brand even when the user never opens your app.
• Hand-offs: content moves between apps.

The words matter too. Apple says the titles and descriptions developers write for each action help Siri understand requests. We'll come back to that, because it's where marketing teams can help most.

Siri AI app skills: playbooks in plain text

Single actions handle simple asks. Skills handle the rest. A skill is a short Markdown file that teaches Siri how to chain several actions for one bigger request.

Each skill has a name of up to 64 characters, a description of up to 1,024 characters and a body of up to 10,000 characters, PPC Land reported. Siri reads the description to decide whether the skill applies. It only reads the body after a match.

A shopping skill might search, add to cart, pick payment and delivery, show the cart and then confirm. The body can also say which steps to avoid and when Siri should ask a question first.

For marketers, this is familiar ground. A skill is basically a written customer journey. Your team already knows which steps confuse buyers and which questions they ask, so that knowledge belongs in the file.

When people must say your app's name

Here's a detail that matters a lot for brands. Siri AI doesn't always need to hear your app's name, but it often does.

So a ride app with a live trip can take "change my pickup to the hotel" with no name, Unite.AI noted. But a cold request like "reorder my coffee pods" needs "in your app name" to reach you.

That makes brand recall more valuable, not less. People will name the apps they remember. A short, easy-to-say app name helps.

What Siri AI can't do yet

Before anyone redraws a channel plan, look at the limits. They're real.

• No App Store use yet. The features don't work in App Store versions of apps, even if the app includes the new code.
• No outside testers. Only development builds and internal TestFlight can use the preview.
• No EU on iPhone or iPad. Siri AI isn't offered on iOS, iPadOS or watchOS in the EU, and Apple only says it's working with the EU on a possible path to bring it there.
• Six languages so far. Vietnamese and Traditional Chinese are listed as coming soon, after English, French, Japanese, Korean, Portuguese and Spanish.
• Daily usage limits apply to server-side features, including Siri AI, and Apple says expanded access will cost a fee later.
• No terms yet. Apple will share the terms and usage policies for these features later, and the API may change during the preview.

Note what's missing from Apple's post, too. No paid placement, no ranking reports and no marketer dashboard were announced, as ContentGrip pointed out.

Is Siri AI the new front door to apps?

It could become one. Today, people find app features by opening the app, tapping a widget or searching the App Store. Siri AI adds a fourth route: ask, and the assistant picks the app and the action.

That route skips your home screen. It skips your banners, your upsell cards and your in-app search. If Siri can book the table, the user may never see your dessert promo. For voice commerce, it's the missing step.

But don't overstate it. Siri AI needs a recent device, a supported language and, often, your app's name. Apple hasn't shared how many people use it. And we don't know yet how Siri will choose between two apps that can do the same job.

So the fair reading is this. Siri AI is a new door that will sit beside the old ones. How wide it opens depends on habits that haven't formed yet.

What Siri AI means for marketers

The closest parallel is AI search. When people started asking ChatGPT or Gemini for advice, brands had to learn how those tools pick sources. We cover that shift in what is GEO.

Siri AI moves the same idea from answers to actions. The question changes from "does the assistant mention us?" to "can the assistant get the job done with us?"

Three things follow from that:

• Your app's actions become a channel. Someone has to own which ones exist and how they're described.
• Repeat tasks matter most, because reorders, order checks and bookings are the jobs people already do on autopilot and will hand to Siri first.
• Measurement shifts. Clicks and screen views won't show what happened inside a Siri chat, so completed tasks become the number to watch, and our guide to AI agent monitoring covers similar metrics.

Your action names are the new ad copy

Siri reads the names and descriptions your team writes for each action. It uses them to decide whether your app fits a request. That's copywriting, even if it lives in code.

Good action copy is plain and specific. "Reorder a past purchase" beats "Quick buy." "Check the status of an order" beats "Track." Write the way customers actually talk, not the way your menu labels read.

Bring your customer service logs to this work. The phrases people type into support chats are the phrases they'll say to Siri.

Pick your first three actions

Don't expose everything. Start with a few tasks that people repeat, that are low risk and that end in a clear result. Here are good first picks by industry:

The "hold off" column is about trust. A wrong booking is annoying. A wrong payment turns into a complaint, a refund and maybe a lost customer who tells friends, so add money actions only once the simple ones work.

How to measure assistant traffic

Apple hasn't announced any reports for brands, so you'll need your own numbers, logged from inside the app from the very first day of testing. Track these:

• Siri starts, per action, per week.
• Completed versus dropped actions, and the exact step where people gave up before the task was done.
• Repeat use.
• Hand-offs: how often Siri sends people into the full app to finish.
• Support tickets that mention Siri.

Apple also offers testing tools. The Siri Integration Advisor in Xcode and a beta Siri Evaluations framework let developers script requests and measure how Siri handles them. Ask your app team to share those results with marketing, not only engineering.

A worked example: a restaurant group

Picture a group with 20 restaurants and its own booking app. The marketing lead and one iOS developer spend two weeks on the preview. This is an illustration, not a real client.

• Week one: the developer exposes three actions: book a table, check a booking and join a waitlist.
• Marketing rewrites the descriptions using phrases from 200 past support chats.
• They write 60 test requests, from "table for four at eight" to "can we move Friday's booking to Saturday."
• Week two: they run the tests and log which ones fail and why.

Say 45 of 60 requests work on the first try. That's 75%, our math on made-up numbers. The failures cluster around vague dates, so the team adds clearer wording to the descriptions and a skill that asks for the date first.

Two weeks of work. No ad spend. When submissions open in the spring, they ship a tested set of actions instead of a rushed one.

No app? What web-only brands can do

This preview is about apps. If your brand has no iOS app, there's nothing to plug in yet. That's fine.

Siri AI still answers questions, and people still ask assistants which brand to pick before they act. Your website, reviews and third-party mentions shape those answers, as they do in other AI tools.

So keep working on the sources assistants read. Our guide to answer engine optimization covers the basics, and our piece on agentic commerce looks at how buying through agents is taking shape.

If you do have an app but no iOS team, talk to your agency or vendor now. The spring window will fill up fast.

Siri AI timeline: what to watch next

Here's how the dates line up so far:

Three things to watch: the terms and any fees Apple sets for these features, whether Siri ever shows paid or ranked suggestions, and when Vietnamese support arrives. The iOS 27 launch coverage is a useful baseline for how fast Apple has moved so far.

Siri AI: our take

Siri AI opening to every app is a real shift. For years, Siri could only do what Apple scripted. Now any app can teach it new jobs, and that changes who controls the path from a request to a result.

But this is a preview. With no App Store use, no EU on iPhone, no published terms and no usage data, anyone selling "Siri optimization" packages today is ahead of the facts.

The smart move is small and early: pick a few actions, write them well, test them with real customer phrases and measure every result. Then wait for spring. You'll have real results while others start from zero.

How Vanaxity helps

We help brands show up where buyers now ask for help: in AI assistants like ChatGPT, Gemini, Claude and, soon, Siri AI. We check how they describe you. Then we fix the pages, data and copy behind those answers, and we track whether those fixes change how often assistants recommend you.

If assistants are becoming part of how your customers find and buy, explore our services or browse more insights.

Frequently asked questions