Apple is working on creating a proactive artificial intelligence-based agent that will be able to autonomously perform multi-step tasks on behalf of the user, including writing code and managing applications.

Bloomberg reports this, citing fresh job openings posted on Apple's official website. In particular, the company is looking for a senior engineering manager who will lead the development of an AI agent.

"Imagine an artificial intelligence that can use a computer just like an experienced person does. It browses web pages, works with various applications, executes commands, writes and debugs code, calls services and chooses the right tool for each step," the Apple job description says.

The company places a separate emphasis on security and confidentiality. The system architecture is planned to incorporate access control models and sandboxing technology so that agents can safely work with user credentials without the risk of information leaking to external networks. The release of the AI agent is expected at the end of 2027.

In parallel with developing its own solution, Apple is making access to the system more difficult for third-party AI tools. Additional security warning screens have appeared in the macOS 27.3 beta. Now, to allow third-party agents (such as ChatGPT) to control a Mac computer, the user will have to go through 14 setup steps instead of the previous four. In the AI industry, this is seen as an attempt to slow down competitors.

While the full-fledged agent is at the initial stage of development, Apple is preparing an intermediate iOS 27.3 update that will expand Siri's integration capabilities with third-party apps.