Apple accelerates Siri to compete with AI agents

In this article you will see:

• What Apple plans to change in Siri in the coming months
• Which agent capabilities the company wants to achieve
• How Siri compares to Dots, Muse and Grok today
• Frequently asked questions about the more agentic Siri

What changes in Siri soon

Two significant improvements are planned for the coming months. The first gives the assistant fine control over Apple's own apps, allowing actions like rotating and cropping images in the Photos app and organizing tabs in Safari. These capabilities should arrive with iOS 27.3.

The second and broader one integrates Siri with third party apps, likely in the spring. This would allow the assistant to place an order via Grubhub or reserve a table via OpenTable. The company treats this as the start of the ability to execute continuous tasks, not just one off commands.

In the long term, Apple wants Siri to track goals, determine the next step and keep pursuing results without the user initiating each step individually. This is what the agents from OpenAI, Dots, from Meta, Muse, and Grok already do today.

Why Apple is accelerating

Apple internally recognizes that it is not keeping pace with rivals in the agent space. Federighi delivered a functional Siri AI version in a very short timeframe, but the company has no illusions that it is at the same level as the competition.

Pressure is coming from the top. Ternus wants Siri to become an agent capable of competing directly with the systems from OpenAI and Meta. The reason is commercial: autonomous third party agents can manage workflows, track multi party goals and orchestrate actions across distinct environments. If Siri remains limited to query responses and app launching, Apple risks ceding high value engagement to agent ecosystems that bypass platform services.

How Siri compares today

Competing agents maintain state across complex workflows, determine intermediate dependencies and keep pursuing user goals automatically. Siri today operates in a question and answer model, with limited automation capabilities.

Apple is restructuring its software pipeline to evaluate an initial request, build an execution plan in system frameworks or external APIs, and execute sequential steps without pausing for incremental user validation at each boundary. For native apps, the system connects voice intent directly to app state. For third party apps, the architecture connects to external endpoints.

The company believes it can make Siri more agent like already next year, not that it has much choice, given current consumer and market demands. The timeline is tight and execution will depend on Federighi's ability to deliver without compromising quality.

What this means for the agent market

Apple's move comes at a time when agentic assistants are becoming the main interface between users and software. OpenAI, Meta and xAI already offer agents that maintain state, pursue goals and execute actions across multiple apps. Apple, with its installed base of billions of devices, has a distribution advantage, but needs to move faster to stay relevant.

For Apple, the risk is that third-party agents become users' preferred interaction layer, reducing reliance on Siri and native services. For developers, the opportunity is to expose functionality via hooks that Siri can trigger, turning apps into tools for agents.

This move connects with broader trends covered on the blog, such as agent orchestration on bridgecode and comparison between Codex, Copilot and Claude Code in VS Code.