The John Ternus Apple AI strategy arrived the way a new chief executive’s first real move usually arrives. It arrived not as a product launch, but as a private conversation that leaked. According to Bloomberg’s Mark Gurman, Ternus held a closed-door meeting with software chief Craig Federighi in recent weeks. He told him that improving Siri is no longer enough. Ternus, who became Apple’s chief executive on September 1, is reported to be frustrated. He sees a reluctance to embrace AI with the same urgency as competitors. He pressed Federighi to accelerate the company’s roadmap and move beyond Siri fixes.

Apple has not publicly confirmed the meeting or the account of what was said. Every claim in this article about that conversation comes from Gurman’s reporting, which rests on unnamed sources. That matters, because the story is not really about a voice assistant. It is about what a new boss does in his first two months. It is about whether the machine he inherited can change speed.

Ternus took over from Tim Cook after a 15-year run that, according to the Motley Fool, carried Apple from a market value of about $350 billion in 2011 to about $4.6 trillion. Cook stayed on as executive chairman. Ternus unveiled the iPhone Duo eight days after taking the job. By mid-October he was on the cover of TIME. The Gurman report is the first public sign of how he intends to run the company he inherited. The sign points at pace.

The John Ternus Apple AI strategy: urgency from the top

The push targets the gap between Apple’s software planning and the speed of the AI race. Gurman reports that Federighi’s organisation plans roughly 12 to 18 months ahead. Apple’s hardware teams plan products several years into the future. That difference was manageable when software shipped on an annual cycle and the roadmap was mostly refinements. It is a problem when the competition ships new AI capabilities in weeks. The product you are chasing keeps moving.

Federighi is not a newcomer to this argument. He oversees software engineering and took charge of Apple’s AI efforts after a leadership shake-up last year. The rebuilt Siri AI has since reached customers. It shipped in beta with iOS 27 on September 14. It offers personal context, understanding of what is on screen, and more sophisticated actions inside apps. Five more languages followed with iOS 27.2 in October. On November 3 Apple opened Siri AI to developers through app skills in the iOS 27.3 betas. The work got done. Gurman’s reporting says Ternus wants more than the work getting done. He wants the ambition to change.

That is the distinction founders should steal. A better Siri is a product improvement. A different AI strategy is a leadership decision. Leaders have to declare it from the top before it shows up in any roadmap. Ternus did not ask for a better voice assistant, according to the report. He asked for a different pace, and he asked it of the person who sets the pace for software.

The 12 to 18 month problem

Short planning horizons quietly kill long bets. An AI assistant that can follow a goal across apps needs sustained coordination. It must remember context and act without being asked at every step, across devices, operating systems and services. It needs the software team, the silicon team and the services team to agree. They must agree on something that will not ship for two or three years. They have to keep agreeing while the market moves underneath them. A team that plans 18 months out can build the next version of Siri. It struggles to build the thing after Siri.

Gurman reports that Apple is exploring exactly that next thing. It is a Siri that behaves more like an autonomous agent. It could follow a goal and make progress on tasks without a separate instruction at every step. He believes Apple could make Siri more like an agent as early as next year. Apple has not announced it, and there is no confirmed release date or final list of features. On the November 2 earnings call, Ternus called agentic AI capabilities “promising” and said Siri is “primarily request-based today” but built on “a modern orchestrator-driven architecture that gives us real room to expand its capabilities over time.” Those are public remarks, not a product announcement. They point in the same direction as the reported meeting. The company knows where Siri has to go. The question is how fast it gets there.

Buying engineers more time

There is also a practical change under way. Gurman reports that the first iOS 28 development milestone gives engineers 50 percent more time to code and explore. It gives them less time on integration, as coding agents have improved. That is a separate part of Apple’s software strategy, and it does not establish which AI features will ship. But it is consistent with a company trying to buy its engineers more room to build ambitious things. That is what a longer planning horizon would demand.

The friction underneath

The report also describes friction around Federighi’s management approach. He reportedly takes a close interest in technical decisions and debates proposals in detail. Some employees believe that style leaves limited room for other voices to influence the direction of Apple’s products. That account rests on the report’s sources, not on a public assessment by Apple. Read it as atmosphere rather than verdict. But atmosphere matters inside a software organisation. The people who have to build the ambitious roadmap are the same people whose voices need room.

Gurman also describes Federighi as an AI skeptic. He has concerns about the technology’s privacy, security, and societal and employment implications. Those are serious concerns, and they are not wrong. They are also the kind of concerns that slow a company down when its rivals are moving fast. That tension is precisely what Ternus is reported to be pressing on. A chief executive does not need his software chief to stop caring about privacy. He needs him to care about it at the speed the race demands.

The personnel picture is wider than one meeting. Mike Rockwell is Apple’s vice president of Siri engineering and the creator of Vision Pro. He took over Siri in March 2025. According to Gurman, he continues to seriously consider stepping down from his role or leaving Apple altogether next year. Bloomberg previously reported in April that Rockwell had reservations about reporting to Federighi. Apple has announced no departure. There is also no indication in the reporting that Federighi is preparing to leave. This article does not claim that he is. What the reporting shows is a software organisation under real pressure for the first time in years. Its most senior people are weighing their futures while the new chief executive asks for more speed.

What founders should copy

Copy the move, not the meeting. A new leader’s first test is whether he can reset the tempo of a machine built by his predecessor. Ternus chose to do it through his software chief, in plain language, behind closed doors. The leak made it public. That is its own lesson: in a company that size, a private message to one lieutenant is a public message to everyone.

The founder version of this is smaller but identical. If your planning horizon is shorter than your competitor’s ambition, you are already behind. No amount of polish on the current product closes it. You have to declare urgency from the top. Do it in words your own team would recognise, before it shows up in the roadmap. Ternus did not ask for a better Siri. He asked for a different ambition, and he asked the person whose organisation has to deliver it.

There is a second lesson in the other direction. Federighi’s caution about privacy and the societal effects of AI is not a flaw in the story. It is the counterweight every AI strategy needs, and a chief executive who bulldozes it will regret it. The test for Ternus is not whether he can make Apple move faster. It is whether he can do it without losing the judgement that made its software trustworthy in the first place.

What to watch next

Three things turn this from a leak into a leadership case study. First, whether Apple confirms or responds to the report at all. Silence would be normal for Apple, and it would leave Gurman’s account as the only version on the record. Second, whether Rockwell stays or goes. A departure next year would confirm that the pressure inside the software organisation is real. It would leave a gap at the top of the Siri team at the worst possible moment. Third, whether an autonomous Siri agent actually appears, and when. Gurman’s “as early as next year” is a belief, not a date. The first real test of Ternus’s urgency will be whether that agent ships while the debate is still live.

Cook built Apple into a machine that never misses a quarter. Ternus has to decide whether that machine can also move fast. His first answer is already on the record, at least the one the world gets to see. Fixing Siri is not a strategy. What he builds instead is the story of his first year.

More CEO stories are in the CEOs archive. Emma Walmsley’s appointment as HP chief executive is the other new-CEO story of the month. An outsider is taking over a hardware giant. The Datavault AI founder who sued his own board shows what happens when a founder and a board disagree. It is a disagreement about the pace of change.