Microsoft’s Bing team did not release its usual weekly development update on its Bing Chat chatbot service on Friday afternoon via the Bing blog. However, that doesn’t mean the team has been taking it easy. The team has been quietly testing a major new feature with a small portion of Bing Chat users for the past couple of weeks.

That new feature is image recognition, and a few Bing Chat users have been given access to it and reporting on the new addition online via Twitter. The feature lets users upload an image, and then ask Bing Chat questions about it.

One of these early testers is Karsten Lehmann, who showed on Twitter how he used it to identify a specific ancient Egyptian temple.

Another Twitter user, known as “@artificialguybr”, scanned an image of a math equation and asked Bing Chat to identify it, which it did as the “Schrödinger equation” along with some more info.

Mikhail Parakhin, Microsoft’s head of Advertising and Web Services, has confirmed that the new feature is currently being flighted to about 10 percent of all Bing Chat users at the moment. It’s also only available for desktop users, but the plan is to expand the support to Bing mobile users as well. Parakhin stated that it should be available “for everyone in a few weeks.”

Parakhin also mentioned on Twitter that Bing Chat support for arbitrary-length document scanning will begin testing next week, again for about 10 percent of all users. He also stated that experimental testing for including Bing Chat in third-party browsers was temporarily pulled because the team “Found a bunch of bugs (especially in Safari).” Testing for this long-awaited feature will be back “soon”.