What Apple Intelligence Features Can You Add to Your iOS App with iOS 27?

Developing an iOS app for iOS 27 now means adding Apple Intelligence features as additions rather than rebuilds: Writing Tools, Siri actions, Image Playground, visual intelligence, and contextual suggestions across Messages, Mail, and Calendar. Announced at WWDC 2026 and shipping in September 2026, most features run on-device through Foundation Models at roughly 30 tokens per second with zero cost per request, and require iPhone 15 Pro or iPhone 16 and later. SiriKit was deprecated, leaving App Intents as the only path for Siri to call into a third-party app, with a two-to-three-year support window. Siri AI launches unavailable in the EU and China pending regulatory approval.

Key Takeaways

• iOS 27 runs on iPhone 11 and newer, but Apple Intelligence features require iPhone 15 Pro or iPhone 16 and later (Macworld, 2026).
• On-device Foundation Models run at about 30 tokens per second on iPhone 15 Pro at zero API cost per request (TechTimes, 2026).
• Apple deprecated SiriKit at WWDC 2026; App Intents is now the only way Siri can call into a third-party app, with a two-to-three-year support window.
• iOS 27 developer beta opened June 8, 2026; the public release ships September 2026, alongside the iPhone 18.
• Many Apple Intelligence features work as additions to a running app, not rebuilds.

 

What new Apple Intelligence features can you add to your existing iOS app?

iOS 27 Apple Intelligence expands across eight categories relevant to third-party apps: writing tools, image generation, Siri actions, visual intelligence, smart suggestions in communication apps, agentic automation, on-screen awareness, and photorealistic image editing. Apple’s own newsroom breakdown of the new capabilities confirms each one ships with the iOS 27 cycle. The sections below cover the features most apps can adopt with the least work.

Writing tools and smart text: the smallest lift, the widest reach

Apple’s Writing Tools — tone rewriting, proofreading, draft generation, and feedback on selected text — work system-wide in any text field on iOS 27, with no integration required from the app developer. A travel app, a healthcare intake form, or a retail support chat gains AI-assisted writing in every field where users type, without a single code change.

A new relationship-aware tone feature adapts message style based on who the recipient is. These features run on-device on every Apple Intelligence-capable iPhone, at no cost per use. For a product manager, this is the rare AI feature that ships to users the moment they update, because the operating system supplies it.

Siri actions, visual intelligence, and image generation

The rebuilt Siri AI can take actions across apps in a single request — finding a photo, setting a reminder, drafting a message, and getting directions — using any app whose content it can see or that exposes actions through App Intents. Visual Intelligence lets users point the iPhone camera at objects and get contextual results inside apps: nutritional data in a fitness app, product identification in a retail app, or bill splitting in a finance app.

Image Playground now generates photorealistic images through Private Cloud Compute and is available to any app through a developer API, with SynthID watermarking built in. Daily generation limits apply, and iCloud+ subscribers receive expanded access at no additional charge. An interior-design app could let users picture a room restyled; a greeting-card app could generate custom artwork from a text prompt.

Contextual suggestions across Messages, Mail, Calendar, and Phone

iOS 27 adds AI-powered one-tap reply suggestions in Messages, natural-language event creation in Calendar, smarter search in Mail, and Call Context, which surfaces relevant data from Mail and Messages on-screen during calls to businesses. A logistics app that fires delivery notifications, a healthcare app with appointment scheduling, or a retail app with customer-service threads can each appear in these system flows once the app exposes its content through App Intents.

The user-facing benefit is concrete: fewer taps, fewer app switches, and information arriving where the user already is. These are small enhancements in engineering terms with an outsized effect on perceived responsiveness.

How does Siri AI integrate with existing third-party apps in iOS 27?

Siri AI in iOS 27 connects to a third-party app through four channels: personal context understanding (reading emails, messages, and files with user permission), onscreen awareness (seeing what is displayed in the current app), App Actions (triggering actions the app has registered), and the Spotlight semantic index (finding content the app has indexed). When a user says “show me the last order from Pedro in the shopping app,” Siri searches indexed content and surfaces it without the user opening the app.

Hands-on testing by MacRumors in June 2026 confirmed that Siri AI can execute complex cross-app tasks — giving directions with multiple stops while editing and sharing a photo — in one request, provided the relevant apps have registered their actions. The same testing noted current beta rough edges, including imperfect auto-correction, a reminder that the feature is still maturing ahead of the September release. The takeaway for app teams: the more actions and content your app exposes, the more often Siri reaches users on your behalf.

What Siri AI can do inside your app without any changes

Some Siri AI capabilities apply automatically to apps that already use system features. Apps that send user notifications gain Siri’s ability to reference those notifications as context. Apps that use Now Playing or AlarmKit gain entity-level awareness without new code. These flow from integrations the app already ships, now upgraded by the Siri AI model. The practical implication: audit what system features your app already touches, because some Siri AI benefits arrive for free on update day.

Do you need to rebuild your app to add App Intents?

App Intents is not a rebuild. It is an additive layer: developers describe what their app can do (actions) and what content it holds (entities), and the system — Siri, Spotlight, Shortcuts — can then find and act on those things. App Intents 2.0 in iOS 27 adds streaming responses for long-running actions, multi-turn conversational follow-ups, and on-screen awareness annotations that let Siri resolve references like “the third one” from what is currently visible. Apple’s WWDC26 session on advanced App Intents walks through exactly how an existing app exposes its actions and content.

There is a deadline worth circling. Apple deprecated SiriKit at WWDC 2026 and gave it a two-to-three-year support window, according to TechTimes’ WWDC 2026 developer coverage. Apps still using SiriKit compile with warnings now and will stop working within that window. App Intents is now the only framework through which Siri can call into a third-party app. Migrating from SiriKit to App Intents is the single most time-sensitive item on any iOS 27 roadmap, and it is the kind of focused upgrade an experienced iOS app development company can scope in a single engagement.

Which iPhones support Apple Intelligence in iOS 27?

iOS 27 as a base update runs on iPhone 11 and newer — Apple’s widest compatibility in years, which TechCrunch called available to more users than any iOS release ever. Apple Intelligence features require more capable hardware, and the requirements split into tiers. The table below summarizes them.

Feature tier	Minimum device	Notes
Writing Tools, smart search, basic Siri suggestions	iPhone 15 Pro / Pro Max or iPhone 16	On-device only; no cloud required
Siri AI with personal context and onscreen awareness	iPhone 15 Pro / Pro Max or iPhone 16	Developer beta now; public release September 2026
Advanced on-device model, new Siri voices	iPhone 17 Pro / Pro Max	Requires 12 GB RAM; not on iPhone 17 base or Air
Image Playground photorealistic generation	iPhone 15 Pro or iPhone 16+	Routes to Private Cloud Compute; daily limits apply

Two regional limits matter for planning. Siri AI is unavailable in the EU at launch pending a regulatory pathway, and all new Apple Intelligence features are unavailable in China pending separate approval, per Apple’s WWDC 2026 announcement. If your app has a large European or Chinese user base, gate AI features by region and keep a non-AI fallback path. Knowing the device split early prevents shipping a feature most of your users cannot run.

Does Apple charge developers to use Apple Intelligence features?

No, for the most common cases. The on-device Foundation Models run at approximately 30 tokens per second on iPhone 15 Pro at zero cost per request, according to TechTimes’ 2026 developer coverage. Apple’s position, stated in its developer frameworks announcement, is direct: “Any app can tap into the on-device models that power Apple Intelligence, and the features you build work offline. And it’s all at no cost per request.”

For tasks that exceed on-device capacity and route to Private Cloud Compute, developers enrolled in the App Store Small Business Program — those with fewer than 2 million lifetime first-time downloads — also pay no cloud API cost. Larger apps may incur costs for Private Cloud Compute beyond the free tier; Apple has not published a public pricing schedule for that tier as of June 2026.

This structure differs from third-party AI APIs in a way that affects unit economics directly:

• On-device Apple Foundation Models: zero cost per request, works offline, no per-token billing.
• OpenAI or Anthropic APIs: every request charges the developer’s API key per token, scaling with usage.
• Image Playground: the app pays nothing; users without iCloud+ hit daily generation caps sooner.

For a founder weighing AI features against a tight budget, the on-device path removes the variable cost that makes most AI roadmaps hard to forecast. Teams already using AI tools to build apps will recognize how much this simplifies planning, and an AI developer can map which of your features stay on-device versus route to the cloud.

What is the difference between on-device and cloud processing in Apple Intelligence?

Apple Intelligence operates across three tiers. The first is the on-device Neural Engine: fast, private, available offline, zero cost, running the rebuilt Foundation Model at roughly 30 tokens per second on supported hardware. The second is Private Cloud Compute: Apple Silicon servers that handle requests too large for the on-device model, using a 32K context window with configurable reasoning levels. The third tier, for the most complex Siri AI queries, routes to Google Cloud.

Private Cloud Compute requests are stateless. Apple states they are “used only to execute the request, are not stored, are not accessible to Apple,” and outside security researchers can verify the claim. The third tier runs on a custom 1.2-trillion-parameter Google Gemini model on Nvidia Blackwell B200 chips with hardware-based confidential computing, and Apple’s contract bars Google from training future Gemini models on Siri queries, according to The Neuron’s WWDC 2026 analysis. Apple reportedly pays Google about $1 billion per year for that model access.

Craig Federighi, Apple’s Senior Vice President of Software Engineering, said at WWDC 2026: “We believe privacy in AI is non-negotiable. Data is only used to execute your request, and outside experts can continue to verify this promise at any time.” For an enterprise, healthcare, or finance app handling sensitive data, the on-device and Private Cloud Compute tiers offer a privacy argument no third-party AI API can match, because the data never leaves Apple’s verified infrastructure. That is a selling point worth putting in front of security-conscious buyers.

When is iOS 27 launching, and what should you build first?

iOS 27 developer beta has been available since June 8, 2026. The public beta arrives in July, and the full release ships in September 2026, likely alongside the iPhone 18, per Macworld’s iOS 27 guide. That gives an existing iOS app roughly three months to test and ship Apple Intelligence features before consumers update. Teams that shipped against earlier cycles — for example, those who built apps for iOS 18 or who upgraded their apps to iOS 17 — already know how valuable that head start is.

Quick wins versus flagship bets: a roadmap framework

Sequencing matters more than scope. The list below moves from lowest effort to highest.

1. Start immediately (low effort, high reach): Verify your app’s text fields work cleanly with Writing Tools — no code required, but layout or input-handling bugs can surface. Register your app’s top three to five actions in App Intents now, which also improves Spotlight discoverability. Index your core content entities so Siri can find them in search.
2. Medium-term additions (one to two sprints): Add Call Context metadata if your app handles appointments, orders, or reservations. Add visual intelligence hooks if your product involves physical objects users scan or photograph.
3. Flagship bets (differentiation features): Integrate the Image Playground developer API so users generate custom visuals inside your app. Build a multi-step agentic flow with App Intents streaming responses — a booking app that confirms, adds to calendar, and messages a contact in one Siri request.

The Foundation Models framework underpins the more ambitious bets, and Apple’s WWDC26 session on the framework details its multimodal image input and the new LanguageModel protocol, which lets developers swap between Apple, Claude, and Gemini models without changing downstream code. Foundation Models now accept images directly — UIImage, CGImage, CoreVideo pixel buffers, and file URLs — enabling on-device receipt parsing, visual product search, and barcode reading with no server call. A conversational layer for your app, similar to custom chatbots, can run entirely on-device using this framework.

Agentic features as a differentiation bet

The most attention-grabbing iOS 27 demo was agentic: Apple’s Passwords app became the first first-party app to autonomously navigate websites and update weak or compromised passwords without user intervention, per TechCrunch’s WWDC 2026 recap. The same recap noted iOS 27 performance gains unrelated to AI: photos load 70% faster and AirDrop transfers run 80% faster than iOS 26.

App Intents 2.0 streaming responses make comparable agentic flows possible in third-party apps. This is the same direction we explore in our work on agentic spatial computing — software that acts on a user’s behalf rather than waiting for step-by-step taps. An agentic flow is a larger design and QA investment than a quick win, but it produces the moments that drive ratings and retention. For teams evaluating whether AI belongs on the roadmap at all, how to build a successful iOS app covers the broader fundamentals that any AI feature sits on top of.

Conclusion

iOS 27 turns Apple Intelligence from a small set of standalone tricks into infrastructure any iOS app can build on. The features that matter most for existing apps — Writing Tools, Siri actions through App Intents, Image Playground, and on-device Foundation Models — arrive in September 2026, work on iPhone 15 Pro and newer, and in most cases cost nothing per request. The clearest near-term action is also the most time-sensitive: migrate any SiriKit integration to App Intents before the deprecation window closes.

Frame Sixty, an AR/VR and spatial computing development studio, helps product teams translate Apple platform releases into shipped features. Whether you want a quick win like exposing your app’s actions to Siri or a flagship agentic flow built on the Foundation Models framework, the three-month runway to the public release is the time to scope it. Our background spans native iOS work through an app development company lens and the broader Apple platform shifts we cover in pieces like what’s new in visionOS 26.

If you’d like to explore which iOS 27 Apple Intelligence features fit your product and budget, get in touch with our team at Frame Sixty. We will help you separate the quick wins from the ambitious bets and ship them before your competitors do.