What Apple Is Actually Planning
Apple currently bans vibe coding tools and autonomous agent frameworks from the App Store, citing security and privacy concerns. But demand has been climbing steadily as Google, Microsoft, and Meta deploy agents across their own platforms. Apple's engineering teams are now drafting a system that would clear AI agents for App Store distribution while keeping the guardrails that define the Apple ecosystem intact.
The framework being developed would allow AI agents to perform tasks, make purchases, book services, and navigate software on behalf of users — without requiring a human to tap through every step. The critical difference from existing App Store apps is autonomy: rather than waiting for user input, these agents would act proactively based on goals and permissions the user grants upfront.
Static mobile applications — download, tap, use — have defined the smartphone era since 2007. AI agents are fundamentally different: they are goal-driven, multi-step, and capable of chaining actions across apps. Allowing them into the App Store means Apple is accepting a new category of software that the current review model was not built for.
Why Apple Is Moving Now
The competitive pressure is real
Google has Gemini agents integrated into Android. Microsoft's Copilot agents are embedded across Windows and Office 365. Meta's AI agents are running across WhatsApp, Instagram, and Ray-Ban glasses. Apple, despite controlling the world's most profitable mobile platform, risks looking behind on the most important software trend of the decade. That is not a position Apple's leadership has historically been comfortable with.
Developer demand is building
The developer community has been increasingly vocal about the restrictions on AI tools in the App Store. Autonomous coding assistants, browser-control agents, and workflow automation tools have been caught in Apple's current blanket restrictions on agent-like behaviour. Clearing a defined path for approved agents resolves this frustration while keeping Apple in control of what actually ships.
WWDC 2026 is the natural stage
Apple's Worldwide Developers Conference lands in June — weeks away. WWDC is where Apple traditionally introduces the frameworks developers need to build the next year's generation of apps. An AI agent framework announcement there would give developers six months to build before the next iOS cycle. The timing of this report is unlikely to be coincidental.
The Guardrails Apple Is Reportedly Building
- Explicit permission scoping — Users would define in advance which apps and data an agent can access, with no ability for the agent to expand those permissions independently.
- Sandboxed execution environments — Agents would operate in isolated containers to prevent cross-app data leakage, maintaining the security architecture Apple is known for.
- App Store review for agent logic — Agent behaviour would go through a new category of App Store review, distinct from the existing app review process.
- On-device processing preference — Where possible, Apple is expected to favour processing agent tasks on-device using Apple Intelligence rather than routing data to external servers.
What This Means for the Mobile App Economy
The App Store has operated on a simple model since 2008: developers build static software, users download and interact with it. AI agents break this model at both ends. On the developer side, the product is no longer a fixed experience — it is a set of capabilities and goals the agent pursues. On the user side, interaction is no longer tap-by-tap — it is intent-driven. This is not an incremental update to how mobile apps work. It is the beginning of a different model entirely.
The implications for app discoverability, monetisation, and review are significant. How do you search for an agent that does not have a fixed interface? How do you price something that does work while the user sleeps? How does Apple's 30% commission model apply when an agent makes purchases across multiple third-party apps? These are open questions that Apple's framework will need to address.
Apple opening the App Store to AI agents is the most consequential platform decision they have made since allowing third-party apps in the first place. I expect the initial framework to be tightly restricted — Apple does not move fast and break things. But once the category is approved in principle, the pace of development will be extraordinary. The question for developers is not whether to build for this platform, but how quickly they can get a compliant agent ready before the first wave of approvals ships. For users, the bigger question is trust: are you comfortable with software acting on your behalf while you are not watching? That is a genuinely new relationship with technology.