Meta and OpenAI Both Gave Their AI a Face. Your App Is Next.
In September, Meta launched Muse and OpenAI launched Dots. Both are AI agents built to do real work for people. Both showed up with a character out front instead of a text box.
That's the tell. The biggest names in AI just told every builder the same thing: AI needs a face. So the next question is yours. How do you add an AI avatar to your app without hiring a 3D team?
Here's what those launches actually were, why a face changes retention and revenue, and what it takes to give your product one.
Building something that talks to users? See how Genies works.
What did Meta and OpenAI actually launch?
Muse is Meta's AI agent app. It does tasks for you instead of just answering questions, and the agent shows up as a personified character, not a blank prompt. Within days of its September launch, Muse took the No. 1 spot on the free iPhone chart in the US App Store, passing ChatGPT.
Dots are OpenAI's always-on agents, announced at DevDay on September 29. Each one appears as a small, bubbly cartoon character. You name your first dot and make it your own.
Two of the biggest AI companies in the world could have shipped another text box. Neither did. The character is a trust play: friendly characters make powerful agents feel approachable at the exact moment people are deciding how much of their lives to hand over.
Why did the biggest AI companies put a character out front?
Because a text box doesn't build a relationship, and a relationship is what keeps people around. Putting a character at the center of AI does three things:
Connection. A face, a voice, and real-time reactions make AI feel like someone, not something. People trust what they can see.
Fun. It shifts the experience from prompting to playing. Think AI assistant meets Tamagotchi.
Adoption. The character becomes the brand. It's recognizable, shareable, and the thing people talk about.
None of that depends on having the smartest model. The labs are competing on who is smartest. Nobody is competing on who you'd actually miss. That's the gap Muse and Dots stepped into, and it's open to every builder.
What does a face change for your app?
The two things you care about most: retention and revenue.
Retention. Content gets consumed once. A character gets a relationship. It remembers each user, reacts to them, and picks up where they left off. That turns a download into a habit, because people come back to someone, not to a screen.
Revenue. Characters people feel attached to get invested in. Outfits, accessories, looks, and unlocks give users a reason to spend, and every conversation gives them a reason to spend again. Customization is the purchase. Interaction is what makes it recur.
What does this look like in a real product?
Say you vibe-coded a language-learning app over a few weekends. It works. Lessons load, streaks count, the AI corrects your grammar. Day one feels great. By day seven, the people who downloaded it have drifted off, because nothing in the app knows them.
Now give it a tutor. A stylized 3D character who greets each user by name, remembers they struggled with the past tense on Tuesday, and reacts while they're still mid-sentence. Hit a ten-day streak and the tutor unlocks a new outfit. Miss a day and the tutor is the one who notices.
Same lessons. Same model underneath. A completely different reason to open the app.
Do you need a 3D team to add an AI avatar to your app?
No. That used to take a studio: artists, riggers, engineers, a voice booth. Genies collapsed it into four pillars.
Looks. Start with a photo or a text prompt and get back a stylized, fully rigged 3D character in minutes, in the art style that fits your product.
Brain. Set the personality, backstory, and the topics it leans into or stays away from. It remembers every conversation and grows over time.
Behavior. The character reacts in real time with its own expressions and mannerisms, and a short voice recording gives it a voice with lip sync driven off the audio. This is what makes it feel alive instead of narrated.
Play. Customization, progression, and unlocks plug straight into the character.
Genies is model agnostic, so you keep the LLM you already use. The labs build the brain. You pick the face.
Why not just generate a video avatar?
Because a video of a character isn't a character. Video avatars re-render a person every time they speak. Costs climb with every second of every conversation, the face drifts from session to session, and nothing can react until the clip finishes generating.
Genies are a rig, not a render. Each character is a real-time 3D model that runs on the user's device, so it reacts mid-sentence, stays on model every time, and keeps your costs from scaling with every frame. For the full breakdown, read Why Everyone Is Doing AI Avatars Wrong.
Where does Genies fit?
Genies is the character layer for any product. The Genies SDK puts a real-time AI character inside your app, game, or website. It's live today for games, with iOS, Android, and Web SDKs in early access. It runs light: the Web SDK is 2 to 3 MB and the iOS SDK is 7 to 8 MB.
You can build it two ways, and both can live in the same product:
One character every user talks to. A tutor, coach, brand host, or support guide that remembers each user individually.
A character every user calls their own. A persistent identity each user customizes and grows with.
It's the same system MLB Players Inc., Sanrio, and WEBTOON signed on to bring their characters to life. For the bigger picture, read Personas Are the New Interface for AI.
See the Genies SDK and early access options.
Frequently asked questions
Do I have to switch LLMs to use Genies?
No. Genies is model agnostic. The intelligence is one input. Keep the model you already use, and Genies adds the face, voice, body, and behavior on top of it.
Will an AI avatar slow down my app?
It's built not to. The character renders on the user's device, and the server sends lightweight animation signals instead of streamed video. The Web SDK is 2 to 3 MB.
Can the character match my app's art style?
Yes. Genies can be built in a range of styles, from stylized humans to cartoon animals to anime, so the character looks like it belongs in your product instead of pasted on top of it.
Is this only for games?
No. Builders use Genies for tutors, coaches, support guides, brand hosts, and story characters. If your product talks to people, it can have a Genie.
Key takeaways
Meta and OpenAI put a character in front of their AI because people connect with something they can see. For builders, that face is a retention and revenue lever, not decoration. You don't need a 3D team or a new model to add one. You need a real-time character layer.
Your app is next
Muse and Dots proved the point at the top of the market. The same move is open to every builder now. Tell us what you're building and we'll show you what your app looks like with a face.