NEWS

Apple contrata engenheiros para construir seu próprio agente de IA

Apple contrata engenheiros para construir seu próprio agente de IA
Image: Redação iMasters
News
Reading
4 min
Tables
1
In short

A job listing identified by Bloomberg shows that Apple is looking for an engineer to lead an agent capable of browsing the web, operating across apps, and even writing code, entering the race against Meta, OpenAI, Anthropic, and Google.

Apple posted a job listing seeking an engineer to lead the creation of its own artificial intelligence agent, capable of browsing the web, operating across applications, writing code, and calling external services. The listing was identified by Bloomberg and reported by The Verge on November 19, 2026.

The listing does not mention a product name or a launch date. But the text of the job posting itself describes quite clearly the type of system the company wants to build, and it is more ambitious than the current Siri.

What the listing describes

The text of Apple's job listing is direct about the agent's goal: to mimic the way a person already works, switching between several tools during a task.

People already do their most valuable work by moving fluidly between tools: a browser, a document, a terminal, an editor, a service (...) We're building AI agents that can do the same… These agents plan and carry out multi-step tasks across apps, browsers, command lines, code, and APIs, and they can coordinate specialized sub-agents along the way.

Excerpt from Apple's job listing, via Bloomberg

According to the listing, the person hired will work alongside teams from three internal Apple divisions: Siri, UI Frameworks, and Apple Foundation Models. This combination suggests that the agent is not an isolated research project, but something designed to integrate with the system interface and the AI models Apple already develops internally.

A race that already has four competitors running

If Apple launches this agent, it enters a race that four other tech giants have already begun. Meta, OpenAI, Anthropic, and Google each have their own version of an agent capable of taking control of a computer and carrying out tasks on its own.

CompanyProductWhat it already does
MetaMuseAccording to The Verge, it already controls the Mac through its own app
OpenAIChatGPT / DotsAccording to The Verge, ChatGPT already controls the Mac through its own app; Dots is mentioned as an OpenAI product in this space, without specific details on how it works
AnthropicDuetMentioned by The Verge as a competitor in this space, without details on how it controls the Mac
GoogleTasks (Gemini)Recently started controlling the Mac, according to The Verge
Appleno public nameOpen job listing

In other words, Apple is not inaugurating the category: it is playing catch-up. Perplexity, ChatGPT, and Meta's Muse already offer this type of screen control in their Mac apps, and Google's Gemini recently started doing the same, according to The Verge.

Siri has already taken the first steps

Apple is not starting from scratch. In September 2026, the company began rolling out an AI-powered version of Siri that allows users to ask about what is on the device's screen and carry out simple actions across applications.

A Bloomberg report, cited by The Verge, suggests that the next steps in this evolution include features like rotating and cropping images within Photos or automatically organizing open tabs in Safari. These are specific tasks, but they already assume Siri making decisions and carrying out actions without an explicit step-by-step command, a central feature of an agent.

The engineering job listing appears to be the next stage of this strategy: instead of isolated features added to Siri, an agent system that coordinates multiple steps and multiple sub-agents at once.

What changes for those who develop for the Apple platform

The most concrete news for those building apps today is not the listing itself, but what The Verge reports about the next system update: according to the publication, with iOS 27.3 Apple would be giving developers the ability to integrate Siri's AI into "all apps," in a feature that would be running as a Developer Preview and that should gradually spread to a larger number of applications.

In practice, if this information is confirmed, it would mean that third-party apps could expose actions to Siri more broadly than today, when integration tends to be restricted to a few specific frameworks and intents. If the agent mentioned in the listing comes to materialize, this app integration infrastructure is probably the foundation on which it will operate: an agent can only "work across apps and call services," as the listing text puts it, if the apps themselves expose hooks for that.

For those who develop for iOS, iPadOS, or macOS, the takeaway is: it's worth following the iOS 27.3 Developer Preview closely. It is there, and not in the announcement of a final product, that Apple would be designing the contract between third-party apps and any agent that comes to coordinate them.

What remains open

The job listing does not include a product name, timeline, or final scope for the agent. There is also no confirmation that it will be launched as a public Siri feature, as a separate product, or as internal infrastructure used by other Apple teams.

It's also unclear whether the agent will run locally on the device, as part of the private computing strategy that Apple has championed for its AI features, or whether it will depend on cloud processing, like the competitors mentioned. These details tend to appear only when Apple decides to officially announce the project, which, as of the original publication of the story, had not yet happened.

Translated from the Brazilian Portuguese original · Read the original