Wonderful Engineering

This Is What The New Facebook M Assistant For Messenger Can Do For You

Image Credit: Facebook

Facebook is coming out with an amazing feature for their Messenger app that can be put at par with an obvious attempt at incorporating artificial intelligence into the mainstream technology. More than 1 billion users will get Facebook’s M assistant on their messenger apps this Thursday, and it sure presents an exciting prospect.

The multi-billion company is now launching the feature of sending suggested responses in the Messenger app, which means that it will give relevant “actions” by judging the context of your conversations to make them “more engaging and personalizable.”

The update will be rolled in iOS and Android, but initially only to the users in the USA.

Image Credit: Facebook

The M assistant was introduced back in 2015 with David Marcus, Facebook’s vice president of messaging products claiming,

“Unlike other AI-based services in the market, M can actually complete tasks on your behalf.”

Two years later, the Facebook M Assistant will now be able to:

Send stickers
Pay or request money
Share your location
Make plans
Start a poll within group conversations
Arrange for a ride using Lyft or Uber

While this might seem something akin to what Google Assistant does with its Smart Replies, Facebook claims that its technology is superior as it can provide more contextual actions as it learns from your history.

So if you are looking to respond to a happy news with a sticker, M assistant will look at the context, your location, and other data to suggest the stickers and save you the trouble of scrolling down the list.

Image Credit: Facebook

This can be applied to everything from sharing your location, sending photos, voice messages, etc.


“It does a bunch of stuff. First, it’s very contextual and what it does, it makes my conversation more engaging and it’s more personalizable,” explained Stan Chudnovsky, Facebook Messenger’s head of product.

And in case M assistant is not able to completely assess the context, it will not suggest anything. Chudnovsky said that this is part of M’s algorithm and training, that its intelligence will make it not show anything in case it doesn’t understand it.

“The original M was a testbed for a lot of things and will continue to be a testbed,” Chudnovsky said. “We get a lot of inspiration from that. Once we learn [something], we’ll take certain pieces that we’ll expand to more people in a contextual way.”

Thursday couldn’t get any slower now, could it?