Microsoft touts Cortana as the best virtual assistant out there, and a Reuters report this morning says Windows consumers won’t be the only ones who can use the service. Per the report, the company is planning to make Cortana a standalone app that will be available on Android and iOS devices. Furthermore, Microsoft is planning to release in the fall a more advanced version of Cortana, one that incorporates machine learning capabilities.

“This kind of technology, which can read and understand email, will play a central role in the next roll out of Cortana, which we are working on now for the fall time frame,” Eric Horvitz, managing director of Microsoft Research, told Reuters. The project is called “Einstein.”

Microsoft didn’t confirm or deny the report, and it declined 425 Business‘ request for comment.

Since it debuted on Windows phones a year ago, Cortana’s presence on Microsoft products has increased. Desktops running Windows 10 will have Cortana built-in, and the virtual assistant could soon be available to a lion’s share of the smartphone market, of which Windows Phone holds about 3 percent.

Making Cortana a standalone product reveals a deeper look into Microsoft’s plan for the service. It’s now clear Cortana is viewed as its own entity rather than a feature of Windows. Making Cortana available on Android and iOS devices fits with CEO Satya Nadella’s plan to get Microsoft products on all devices running all operating systems.

More important is the revelation that Microsoft intends Cortana to be a machine-learning tool. Virtual assistants like Cortana, Apple’s Siri, and Google Now are best recognized by their voice recognition and response capabilities, but Microsoft wants Cortana to become a predictive agent, something it and Google Now already do to a limited degree.

The concept of machine learning is fast gaining traction and importance in tech engineering houses. In the near term, machine learning projects like the proposed Cortana can scan through applications on a device or cloud network, gather data, and interpret that data to administer centralized notifications or even make predictions. But user experiences such as that are early tests of larger-scale machine learning capabilities that will be required to handle big data assignments for companies.