Meta AI: Social Twist on AI Assistants

This post is part of my “Exploring the Landscape of AI Assistants” series — see all the assistants I’ve reviewed.

Of course, with all the buzz around AI and AI assistants, I am sure Meta felt FOMO and jumped in. This week I explored, Meta AI, an AI assistant from Meta, on my iPhone. I will be comparing it with Anthropic’s Claude, X’s Grok, Google’s Gemini and OpenAI’s ChatGPT, which I recently examined.

Note:

You can explore the key user flows for Gemini, ChatGPT, Grok, and Claude on Meeako

Connected World

Meta’s vision has always been about building product and services that help people connect with friends, family and the world. It’s evident that Meta AI has the same vision. ~~I am looking forward to see how they’ll build social features around an AI assistant.~~ (See Sharing, Posting to Feed, Animating Images, Editing a video, Remixing a prompt )

Soft Launch

I haven’t studied other companies such as Google and Microsoft – both of which have an ecosystem of products & services – but I am pretty sure they followed a similar path. I noticed Meta, before launching a dedicated, standalone app for its AI assistant to compete with the likes of ChatGPT, Grok and Claude etc, began slipping AI features into existing products such as Instagram, WhatsApp and Facebook.

Here’s what I noticed on Instagram, Messenger and WhatsApp:

I guess there are perks to having multiple product with billions of daily active users. I do see some advatanges to Meta:

access to feedback from ~3.5 billions of users.
opportunities to train their model (llama), while offering users more personalized responses and recommendations.
Ability to explore a range of vertical and horizontal use cases.

Positioning

At this point, I am convinced that every app in this space wants to become your “personal” and “everyday” assistant – and Meta AI is no different.

Understands You

However, when it comes to building an AI that understands you, I think Facebook has a head start. Sure, if you use any AI assistant long enough, it will eventually get to know you, perhaps more than you know yourself, but the type of user data Facebook already has – what you like, your friends, your groups, even your dating activity – could give it an early advantage. Of course if no one ends up using it, then that’s a whole different issue.

I like that Facebook is highlighting this as a key differentiator in their messaging.

Stay Connected

I know that, unlike Google and Amazon, Facebook doesn’t have a dedicated assistant device – like Google’s Google Assistant and Amazon’s Alexa. But it seems Facebook is using their Ray-Ban’s AI glasses to support their ‘Stay Connected’ vision.

Sign Up

Meta has one of the longest sign up flow – it’s painfully long. Not sure, why. In today’s day, going through 13-14 screens and user interactions feels like a sin. But I guess, when half the world’s population is using your product, optimizing your sign up flow may not be a top priority.

Edit a Video

This is a paid feature [at least] in Google Gemini, so I am happy to see Meta AI allowing users to test for a limited time.

Demo

Creating, editing and sharing videos and images are more core to Meta than any other competitors I have reviewed so far, therefore, they will have to be ahead of the pack.

Free for limited time

It’s a free feature for now, but they plan to make paid later. It’ll be interesting to see how that plays out, as there are some dedicated video and image AI assistants already in the market (e.g. OpenAI’s Sora etc)

Creating and Animating Images

Create Images

At this point, other than Claude, every AI assistant I have reviewed generates images – so no surprise there. However, I will say, the quality of images produced by Meta AI is the best I’ve seen so far.

Animate Images

Now, this is absolutely new for me – so far, I have not see any other assistant do this. Meta AI lets you create GIFs from the images it creates – pretty neat!

I wonder, if this feature is a part of Meta AI’s social angle i.e. allowing users to bring their memories to life, so that they are more likely to share them with friends and family?

Mind of its own

I asked Meta AI to animate the image of my cat. It created a realistic image of a cat that resembled like mine, however, it did not actually animate it. So, when I asked why, it gave me a bunch of images without any explanation. I tried clarifying again but got a similar response. To me, it shows Meta AI is not quite there yet, but I am certain it’s only a matter of time.

Sharing Made Obvious

As both a user-generated content platform and a social media company, Meta cares not only about users creating content but also about sharing it. This is clear in Meta AI, where the share button sits prominently in the top-right corner of the chat instead of being buried in a menu.

Posting to Feed

Here we can see social angle in action: you can share your chats to a feed, Discover tab, where other users can engage (like, comment, share or even re-use your prompt)

Remixing a Prompt

This is fun! On the Meta AI you can see prompts and conversations other people had with Meta AI (if they choose to share them publicly). You can remix a prompt you see, which starts a new private conversation with Meta AI based on the prompt you remix.

I can already see a lot of people having fun with this feature; definitely bringing in a lot of engagement – like, comment, share – to the platform.

I see some growth loops here — the better the AI image, the more people want to remix the prompt to create their own version. They then share it, which starts the whole cycle again, while also helping Meta improve its models and share the word.