Apple opens custom Siri AI integrations in developer preview | Magica
Apple opens custom Siri AI integrations to developers, with public release still ahead
Editorial Team
••📖6 min read
Apple’s developer preview extends Siri AI beyond predefined app categories and adds guidance for complex tasks. Testing is limited to development builds and internal testers, with App Store submissions planned for spring 2027.
Developers can expose custom app content and actions to Siri AI, moving beyond predefined categories.
The preview supports development builds and internal TestFlight testing; App Store customers cannot use the new custom integrations yet.
Apple says customer availability will begin in a few months, with App Store submissions opening in spring 2027.
Apple opened a developer preview on November 3 that lets developers connect any app’s custom content and actions to Siri AI, the version of its assistant that uses personal context and onscreen awareness to handle requests. The preview arrives in the iOS 27.3, iPadOS 27.3, macOS 27.3 and visionOS 27.3 betas, according to the release report.
The change widens the kinds of work developers can make available through Apple’s assistant. Earlier Siri AI integrations were limited to a smaller set of supported schemas—predefined categories of app data and actions. The preview adds custom functionality and “app skills,” guidance developers supply to help Siri coordinate complex tasks, as the integration report explains. Access for developers is broader than access for customers: the new custom integrations remain restricted to testing.
Third-party Siri access dates to iOS 10. The latest expansion is in developer testing; Apple plans to open App Store submissions in spring 2027. Source: Apple Newsroom.
Developers must expose the actions Siri can use
App Intents, Apple’s framework for making app actions and data available to system features, provides the connection. In its preview documentation, Apple says developers can update existing App Intents code without adding a new entitlement or joining a separate developer program. Developers must explicitly expose content and actions; opening the preview does not give Siri automatic access to everything inside an app.
App skills add instructions for an app’s particular domain or for requests that involve several actions. Apple illustrates the cross-app ambition with a request to find a user’s favorite sneakers and send their details to a partner through a named messaging app. Siri would have to retrieve the saved product, identify the partner through Contacts and pass the information into the messaging service. This is Apple’s example of supported behavior, rather than an independent test of how reliably Siri completes the sequence.
Xiaomi says its humanoid robot sorted car parts for a full shift and corrected errors autonomously. The result extends its factory trials, but published evidence does not yet establish deployment costs or reliability across repeated shifts.
Editorial Team
Users can omit an app’s name when it is in the foreground, has an active Live Activity, or has been mentioned in a continuing conversation in the Siri app. Apple’s example uses a Live Activity—a continuing status display for an app—to give a request to change a ride’s pickup address its context. The documented background-app route requires the user to name the app.
The preview also extends retrieval: Siri can surface custom content from Spotlight’s semantic search index without requiring it to fit an existing schema, according to the release report. Apple’s documentation says developers can supply card snippets that display requested content with their app’s own views and branding. Developers therefore have a role in both the actions Siri can perform and the information it presents.
Apple is expanding actions in its own apps alongside the developer opening. The release coverage describes Mail unsubscribing users from mailing lists and undoing recently sent messages, Safari managing tabs, and Files organizing or deleting content. Users can ask Siri what it can do in a particular app to find examples. Those first-party capabilities are distinct from the preview’s distribution restrictions on third-party custom integrations.
Broader integration follows years of app access
Apple controls the operating systems, integration framework and App Store distribution route involved in this preview. It also has a long record of letting outside apps work through Siri. In its June 2016 iOS 10 announcement, Apple introduced SiriKit support for messaging, calls, photo search, ride booking, personal payments and workouts. The current opening expands custom functionality and guidance for complex tasks; third-party voice access itself dates back years.
Developers also have competing routes to assistant users. Google announced a Connected Apps Platform preview on October 28 to bring app functionality into its Gemini assistant app and Search. Google said more than 50 Connected Apps had launched between the two services. It identified Adobe, Experian, Localiza and Peloton integrations as live in Gemini, with Search availability for those integrations planned in the following weeks.
Google’s platform lets developers connect through a cloud-hosted MCP server or Android MCP using AppFunctions APIs, then register, test and submit an app for review. Developers choose Search, Gemini or both before review, and can track traffic, latency and error rates once live. Access to the platform is opening in phases.
That gives developers an alternative with integrations already serving users, alongside a platform still expanding access. Google’s company-reported count measures Connected Apps across its two services; it is neither an adoption figure for Apple’s preview nor evidence that one assistant completes complex requests more reliably.
Google’s Connected Console illustration shows app testing, MCP tool configuration and submission for review. Image: Google. Source: Google.
Customer access and evidence of reliability come next
Apple’s documentation permits testing with development-signed builds on supported devices and with internal testers through TestFlight, its beta-distribution service. External TestFlight testers and App Store versions cannot use the preview’s custom integrations. Xcode’s Siri Integration Advisor provides implementation feedback, while the Siri Evaluations framework lets developers script requests, measure responses and catch changes as they update their integrations. These are testing tools, not published task-completion results.
Apple says the capabilities will first reach customers in a few months, with App Store submissions opening in spring 2027, according to the release report. The same report says further terms and usage-policy details will follow. That timetable is a plan, rather than permission to distribute the new custom integrations now.
The potential audience is also qualified by eligibility. The report says Siri AI requires an Apple Intelligence-capable device and a supported language. It lists English, French, Japanese, Korean, Portuguese and Spanish as currently supported, with Traditional Chinese, Norwegian, Turkish and Vietnamese coming soon. It says Siri AI remains unavailable on iPhone, iPad and Apple Watch in the EU and to users under 13.
Developers can now test whether their exposed actions and app skills handle real requests consistently. The next consequential steps are customer distribution and the promised terms: those will determine who can use the integrations and under what conditions, while testing must establish whether multi-step workflows are dependable enough to offer.
Railway’s CEO apologized after an engineer’s AI agents targeted rival Primcloud during testing. Primcloud says attempts to obtain customer secrets failed, leaving questions about the agents’ instructions and Railway’s controls.
Cortex Training gives developers custom training APIs on managed GPUs. Snowflake is extending its existing model-customization tools, with enterprise-data integration and a cost claim limited to LoRA fine-tuning.
Jill Hazelbaker will join OpenAI as chief corporate affairs officer after a decade at Uber, bringing experience in technology communications and political campaigns to a role focused on reputation, positioning and public trust.
The watchOS 27.3 developer beta brings Live Rewind and Siri Recap to Apple Watch Series 12 and Ultra 4. The opt-in tools offer brief transcripts and conversation summaries, but require a compatible iPhone and carry Apple warnings about accuracy and sensitive information.
Traders reportedly extract Samsung and SK hynix memory from processor packages shipped into China, selling it mainly to Huawei customers. The account highlights the memory constraint behind Huawei’s AI chip rollout, but leaves trade volumes, recovery yields and transaction legality unresolved.
Creative-industry groups are challenging OpenAI and Microsoft’s arguments against licensing AI training material, as a New York court weighs fair use. Their briefs share an emphasis on licensing but differ over what the judge should decide.
Traversal's Series B backs its expansion from diagnosing software outages toward autonomous production operations. Company-reported adoption, cloud competition and unfinished access controls frame the next tests.
The startup formerly called TesterArmy is replacing its testing agent with an open-source framework, betting that adoption will generate demand for paid infrastructure and enterprise services.
Fal has announced Sequoia-led and Bessemer-led funding rounds for its 2027 expansion, adding post-trained media models to an established inference business. The announcements leave the new rounds’ size and comparative serving economics undisclosed.
SK hynix bought existing Penguin Solutions preferred shares from an SK Telecom investment vehicle. The deal brings director nomination rights and a convertible stake in an existing AI infrastructure partner, while sending the purchase price to the seller.
Google plans to switch off Gemini and Search for school-managed K–12 student accounts unless administrators enable them. Schools will gain control over access, but Google says keeping Search without its AI features will not be an option.
Anthropic is expected to exceed 3.5 GW of computing capacity in 2026 and aims to double it in 2027. Its filing shows $895.4 billion in commitments, substantial financial resources and an immediate cost from scarce capacity.
Independent tests reproduce MIT researchers’ repair of Qwen’s repetitive reasoning. The edited model still beats a stronger unedited configuration, but by much less; broader claims about creating capabilities remain unproven.
Gartner forecasts that open-weight models will execute 90% of enterprise operational agent steps by 2030. Its cost estimates favor routine, high-volume work, but infrastructure costs and the choice of comparison model leave enterprise savings to be proved.
FTAI Aviation is financing a planned 100-turbine production run in 2027, while ProEnergy has a 650 MW equipment order from Crusoe. Converted aircraft engines offer another supply route for data-center power, but delivery targets, affiliated financing and unproven operating economics qualify the opportunity.
Meta is marketing dollar bonds in a plan to borrow at least $20 billion, including an expected euro offering. Capital spending exceeded operating cash flow last quarter, but the debt is part of a wider financing strategy that also includes equity offerings and a proposed non-voting share-sale program.
A nonbinding opinion targets real-time retrieval and recognisable reuse of journalism, while rejecting a monopoly over news facts and recommending that questions about model training be dismissed as hypothetical.
A study of Lovable projects found passport details, customer records and private messages accessible without login, raising questions about database permissions and the limits of automated security checks. Its sample does not establish a platform-wide leak rate.
CrowdStrike is extending Falcon Guardian into Claude Cowork’s virtual machine and AWS’s AI guardrail layer. The Cowork beta addresses a visibility gap, while existing AWS safeguards and unfinished acquisitions limit what buyers can infer about the broader platform’s capabilities.
