Apple Intelligence: What AI on the Smartphone Means for Enterprise IT

Apple Intelligence brings generative AI directly to iPhone, iPad as well as Mac, and so into the hands of employees who need no IT project to use it. For enterprise IT this means steering usage before it begins: with a classification of data, a check of the device fleet and a limited pilot. The announcement of 10 June 2024 gives enough pointers to start planning now.

This article summarises what Apple announced at its WWDC developer conference, shows which questions enterprise IT should ask and offers a checklist for preparation. This article provides an overview and does not constitute legal advice.

What did Apple announce with Apple Intelligence?

On 10 June 2024 at WWDC, Apple announced an AI system for iOS 18, iPadOS 18 as well as macOS Sequoia that runs language models on the device and, where needed, on Apple's own servers. The features cover writing, images and Siri. The writing tools rewrite, proofread and summarise text in Mail, Notes and other applications. Notifications are sorted by urgency and summarised. According to the announcement, Siri understands personal context, such as messages and appointments, and can carry out actions inside applications.

Apple sets two conditions. The features run only on devices with enough computing power: on the iPhone 15 Pro and iPhone 15 Pro Max as well as on iPad and Mac with the M1 chip or later. In addition, according to Apple they launch in autumn 2024 as a beta, initially in US English, with further languages and features to follow over the course of the next year. Older devices in the fleet remain excluded.

How does Private Cloud Compute work?

Private Cloud Compute is Apple's method of running more demanding requests on servers with its own chips, without the data being stored or seen by Apple. According to Apple's description, the device processes as much as possible itself and sends the servers only the data a given request needs. Apple states that independent experts can inspect the server code. For decision-makers this is a vendor's promise whose effectiveness is shown only by independent review and by documentation.

A third processing level involves ChatGPT. With the user's consent, Siri can pass requests to OpenAI's ChatGPT without an account being required, and the writing tools can use it as well. For IT this means three data paths on one device: local, via Apple's servers and via an external provider. Each path needs its own assessment.

Which questions must enterprise IT settle now?

Enterprise IT must settle five questions: which devices are affected, which data may be processed, whether the features can be controlled, who should use them and how the benefit is measured. The first three are IT tasks, and the last two belong to the business units.

The third question cannot be answered conclusively at the time of the announcement. Apple said nothing about management by enterprises in the announcement, and documentation on it is still to come. Enterprise IT should therefore record in advance what it expects from management: features that can be switched off individually, a separately blockable path to ChatGPT as well as logging of usage. Data protection and information security review each of these requirements case by case, and legal assessment lies with legal specialists.

What changes for employees?

For employees, AI features appear in the everyday tools they already use, in Mail, Notes, Messages as well as Phone. The hurdle of opening a separate AI tool disappears. That lowers the barrier to entry and raises the risk that confidential content is processed unintentionally. A field sales employee who polishes a customer email with the writing tool makes a data decision without noticing.

A simple measure follows from this: a one-page policy that says in examples what is permitted. Two examples per data class work better than a paragraph of principles. How teams handle parallel AI tools in practice and when this turns into sprawl is described in the article on AI subscription sprawl.

Does Apple Intelligence change the competitive position of the vendors?

Apple Intelligence mainly changes the reach of AI features, because they arrive on devices that are widespread in companies. The combination of local processing, own cloud and an external partner resembles the approach of other vendors. In office software Microsoft has taken a comparable route with Microsoft 365 Copilot, where AI is integrated into the applications. For companies this creates a platform decision: which AI features come from the operating system, which from the office suite and which from applications the company builds itself?

Whoever does not take this decision has it taken by the vendors. A simple grid is sensible that assigns each feature to one of the three sources and names owners per source. Preparing for new announcements always follows the same logic, described in the article on the AI hype cycle: announcement, availability and benefit are three different events.

How do you plan a pilot for Apple Intelligence?

You plan a pilot in four steps: choose the group, limit the data, set the metrics and decide after eight weeks. The pilot group comprises ten to fifty people with suitable devices and typical writing and communication tasks, for example in sales, customer service or project coordination. The data rules are binding for the group. What is measured is time per task, quality of results as well as user satisfaction, each against a baseline taken before the pilot.

After eight weeks executive management decides on the basis of the measurement whether the features are released to further groups, under which rules and with what training need. Without a baseline this decision has no evidence behind it, so the measurement comes before the release of the features. How pressure becomes a robust plan is described in the article on AI activism.

AI on the smartphone with Niehues Consulting

Niehues Consulting is a consultancy for AI management and organisation design from a delivery perspective, based in Düsseldorf. Founder Marin Niehues brings AI into operations, in projects ranging from VC-funded scale-ups to DAX 40 corporations and in over 20 countries on six continents. In the service area of AI management the work runs from business analysis and requirements through the business case to the selection and sequencing of applications and portfolio steering.

The approach follows the Niehues Delivery Model. Two of its review questions matter for features such as Apple Intelligence: which business outcome should change measurably? And: which metric shows after 90 days that the solution is landing? Scope and terms of an engagement are discussed in the free initial consultation, which you can arrange via the contact page.