Ready for a Siri AI agent? Apple under pressure to upgrade its assistant faster

Apple CEO John Ternus is pressing the company's software leadership to speed up its artificial intelligence work, according to reporting by Bloomberg's Mark Gurman. Gurman describes Ternus as publicly pleased with the Siri AI launch while privately urging Craig Federighi, Apple's senior vice president of software engineering, to accelerate the roadmap and go beyond incremental fixes to Siri, with a more agentic assistant as the goal. Gurman attributes the friction in part to Federighi's reputation as an AI skeptic.
The current Siri AI is described as a notably improved voice assistant. Gurman points to its ability to automatically fix weak passwords as a modest form of agentic behavior, though he notes that automatically generated passwords bring their own problems. The gap is with agents from AI-first companies that aim to proactively organize tasks, operate a computer or write advanced code. Gurman also characterizes the rival landscape as uneven: some agents, such as OpenClaw and Anthropic's Claude Code, are highly functional, while others, such as Google's Gemini Spark, are not yet there.
The pressure is framed around timing. Apple builds hardware well suited to running local models, yet has moved more slowly than competitors on AI software, while those competitors have shown they can turn promised agent features into shipped products.
Apple's slower pace has arguably helped its reputation as public skepticism about AI grows. Some users are frustrated by software that feels bloated with mediocre or unnecessary AI features, particularly as hardware prices climb during a memory shortage. Apple's privacy-first approach has been a differentiator against other AI products. Gurman cites Meta's Muse agent as an extreme example of data-hungry design, though trade-offs between privacy and usefulness appear in other personal agents too; Gemini, for instance, works best with more of a user's data.
On the other side, Apple spent roughly two years promoting Siri AI before delivering an improved voice assistant rather than a breakthrough product, which may mean users are now ready for more advanced capabilities. Gurman also notes that AI companies are building their own hardware. If such software is paired with devices designed from the outset around assistants that see, remember and act, Apple's long-standing advantages could carry less weight, he wrote.
An agentic Siri would most likely extend features users already have. The current on-screen awareness capability is limited to what is visible at that moment, meaning only the portion of a long email a user has scrolled to. A more capable Siri could work through an entire email on its own and surface the relevant insights.
In the near term, Gurman expects Ternus's push to mostly deliver on commitments Apple has already made. Siri AI still lacks advanced in-app actions, a capability Apple announced in 2024 that would let third-party apps build plug-ins and pathways into Siri and allow the assistant to work across both Apple's own apps and outside ones with more precision. Gurman wrote that Apple began discussing the feature again during the week of the first beta of iOS 27.3.
If Siri became a competitive agent, many users who trust Apple's ecosystem might stop installing apps such as ChatGPT or Claude. Gurman judges that unlikely for now given the distance between Apple's AI and consumer products from OpenAI and Anthropic, though not out of the question as AI backlash and privacy concerns grow. For agents specifically, newer entrants such as OpenAI's Dot and Anthropic's Duet are not yet competitive either; ZDNET's David Gewirtz found both a little odd and not quite ready for prime time.
Privacy questions shadow some rival features. ChatGPT's Computer History, which builds a timeline of app and website activity and is available only on the Mac, drew concern when it launched in August. The feature is opt-in and can be switched off, but OpenAI said it does not encrypt sensitive information the feature encounters, leaving it exposed to attackers. Apple could build more secure versions of such features in-house using its privacy reputation. Still, users who value Siri AI's subtlety may resist anything resembling ChatGPT or Claude unless every feature is opt-in.
Why it matters: Developers building for Apple platforms have a concrete cue in the reappearance of advanced in-app actions during the iOS 27.3 beta, since that work determines whether third-party apps can plug into Siri at all. For product teams, the strategic question is whether Apple's privacy posture and hardware become a stronger draw than the more autonomous agents shipping elsewhere, and my inference is that any shift toward an agentic Siri will be paced by how quickly those app integrations actually arrive. Users stand to gain capability but may also see more AI pushed into everyday interactions, which makes opt-in design the likely flashpoint.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Apple’s public-facing satisfaction with the Siri AI launch obscures internal pressures and tensions.