The AI features Apple unveiled alongside the iPhone 17 lineup in September share a common pattern. Notification summaries, call transcription, real-time conversation analysis through the Apple Watch's Audio Intelligence — none of them work outside their host app. Users encounter AI separately in Notes, Mail, and Phone, each in its own silo. Step outside the app, and whatever context that AI had built up simply disappears.
As long as AI is accessed through individual apps, users are the ones who have to bridge the context lost between them. And the more sophisticated AI becomes within each app, the more glaring the cost of that gap becomes.
The Time Spent Filling the Gaps Between Apps
Track a solo founder's or independent director's day, and the same pattern emerges: they spend more time gathering and organizing the information a decision requires than actually making that decision.
A discussion from Slack gets copied into Notion, a decision in Notion gets written up in an email, and a date from that email gets entered into a calendar. None of this shuttling is automated. Each app's AI holds context only within its own walls; connecting one app to the next is still a human job.
Even if Audio Intelligence summarizes a call in real time, turning that summary into a calendar entry still requires the user to open the calendar app themselves. The Apple Watch has gotten remarkably good at capturing and parsing speech, but if that output doesn't connect automatically to other apps, all that sophistication stops dead at the app's edge.
The gap exists because there's no pathway for apps to exchange context with one another.
The Platform Logic That Keeps the App Layer Intact
There's a clear business logic behind Apple keeping AI confined inside apps.
Apple's Services segment generated more than $100 billion in revenue in fiscal year 2025, a substantial share of it from in-app subscriptions and payment fees. If Apple built platform-level AI agents that bypass apps altogether, it would undercut developers' revenue base and reshape the foundation of that entire ecosystem.
At the same time, keeping apps as the runtime environment for AI is also a way of retaining control over the developer ecosystem. To integrate with Apple Intelligence, developers have to work within the APIs and rules Apple sets. If the app layer disappeared, that control structure would disappear with it.
As a platform operator, this is a rational choice for Apple to make. It just doesn't always align with solving the context loss users run into every day.
As long as AI operates through apps, a human hand will keep being necessary somewhere along this chain.
Connectivity to Check Before You Check the Features
This changes the criteria for choosing an AI tool.
It used to be enough to ask, "What does this app do best for my core work?" Now there's a second question that matters just as much: "How does this tool exchange context with the others I use?"
An app with brilliant features but a hard time exchanging data with the outside world becomes a bottleneck in the overall workflow. A tool with simpler features that shares context smoothly with others speeds up the whole operation instead. That difference is invisible when you're using a single app in isolation — it only shows up cumulatively, across a whole day spent moving between tools.
When evaluating a product or tool, look first at how it fits into the flow of someone's actual behavior, not at how impressive it is in isolation. The more often a tool breaks that flow and forces you to stitch things back together by hand, the lower its real usefulness. That same standard now applies to choosing AI tools.
Where to Consolidate Your Context
Read Apple's announcement in reverse, and it points to one practical question: if each app's AI only holds context within its own domain, who's responsible for building the context that spans across apps?
The most realistic answer is to deliberately design a hub where context accumulates — a single place where every conversation and every decision gets logged, with an AI that can access that record. That way, no matter which apps come and go, the context stays put at the hub.
People who take this approach evaluate a new tool by its openness before its AI polish. They check what formats it can export data in, whether its API is open, and whether other tools can read that data. Even when they use Notion or Obsidian, the goal isn't to run AI inside that app — it's to make sure the context piling up there can be read by other AI systems. The yardstick for choosing tools has shifted from which one has better features to which one connects better.
Because Apple didn't give up its grip on the app layer in this 2026 announcement either, that gap remains the user's problem to solve. The real variable in AI-driven productivity isn't which apps you use — it's how context flows between them.
