Meta solves the problem of agent-based artificial intelligence with the launch of Muse

Meta has been taking steps to develop its AI game in recent months, launching Muse Glimmer and Muse Spark earlier this year. Now it’s time for the company to make its big agency push by introducing the real Muse.

Under the hood, Muse runs on Muse Spark, and as Meta says, Muse is designed to make some work easier. This means that users can submit requests to Muse for anything from composing and sending an email to helping close a deal on a new purchase, and the system will develop its own plan of attack and then proceed to actually do the work once authorized. A separate agent called Sentinel helps make sure Muse doesn’t go online or do anything you don’t approve.

Obviously, carrying out serious tasks like this will require you to put a lot of trust in Muse and allow it access to your accounts, so Meta explains in detail the security steps it takes to keep your data safe, such as pushing it into a personalized virtual machine. Your interaction with this virtual box then occurs through encrypted connections to the Meta apps on your devices.

This security system is due to receive an update, and later this year the Muse Confidential virtual machine will arrive, which will make it so that not even Meta itself can see your data.

Like similar agents, Muse has a memory and must learn your preferences as you express them.

While this all sounds good, it’s the same thing that all the other big AI players are doing, and it remains to be seen whether Meta has enough juice to make Muse the first choice users turn to when choosing an AI to work on their behalf.

If you’d like to try Muse, you can sign up for Muse.ai to gain access. Availability begins in the US with support for iOS and Android. Meta reports that Muse for smart glasses is coming soon.