Tubi just did something interesting. The Fox-owned streaming service launched a native app inside ChatGPT, making it the first major streamer to put its entire catalog directly into OpenAI’s platform. You can now type “@Tubi” and ask for “a thriller for girls’ night” and get recommendations pulled from their 300,000+ title library.
This is genuinely novel. Netflix and Prime Video have been tinkering with AI recommendations within their own walled gardens, but Tubi is going where the users already are. And with ChatGPT hitting 900 million weekly active users, that’s not a small bet.
The Discovery Problem Nobody Solved
I’ve been thinking about streaming discovery for a while now. It’s broken. Completely broken.
Every platform has the same problem: too much content, terrible search, recommendation engines that feel like they’re stuck in 2015. You end up scrolling for 20 minutes, give up, and rewatch The Office for the dozenth time. Tubi has 300,000 titles. Good luck finding anything in that haystack without accidentally landing on a direct-to-video disaster from 2003.
The irony is that Tubi tried this before. They built “Rabbit AI” in 2023, their own in-house ChatGPT-powered recommendation tool. Shut it down a year later. Building AI products is hard. Maintaining them is harder. Competing with OpenAI on their home turf is basically impossible if you’re not Google or Microsoft.
So instead of fighting that battle, Tubi’s doing the smart thing: embedding themselves where people are already asking questions.
Meeting Users Where They Live
This is what interests me most about this move. It’s not about the technology, which is fairly straightforward. OpenAI opened up their app ecosystem back in October, and we’ve seen integrations from Spotify, DoorDash, Booking.com, and others. The technical lift here isn’t revolutionary.
What’s clever is the positioning. ChatGPT has become the default “figure this out for me” interface for millions of people. You ask it about recipes, travel plans, coding problems, gift ideas. Why wouldn’t you ask it what to watch?
Tubi has 100 million monthly users. That sounds impressive until you remember Netflix has 260 million subscribers and Disney+ has over 150 million. Tubi is ad-supported, free, and competing for attention in an incredibly crowded space. They need distribution channels that don’t cost them customer acquisition dollars.
This ChatGPT integration is essentially free marketing. Every time someone asks ChatGPT for movie recommendations and Tubi surfaces in those results, that’s brand exposure they didn’t have to pay for. The friction to actually starting playback is still there (you get links back to Tubi’s platform), but discoverability is half the battle.
The Broader Pattern
What I find fascinating is how this reflects a larger shift in how we think about product distribution. The old model was building everything in-house, owning the entire user journey from discovery to consumption. That’s still what the big players do because they can afford to.
But for everyone else? The calculus is changing. Why build your own AI recommendation system when you can plug into ChatGPT’s existing infrastructure and user base? Why train your own models when OpenAI has already done the heavy lifting?
We’re seeing this pattern repeat across industries. Companies are increasingly treating large language models like platforms rather than competitors. It’s the same dynamic that made the iOS App Store successful: you meet users where they already are instead of trying to drag them somewhere new.
Tubi shut down Rabbit AI because maintaining a competitive AI product requires resources most companies don’t have. The technology moves too fast. User expectations evolve too quickly. Unless you’re betting the company on AI, you’re probably better off integrating with someone who is.
I wonder how long it takes before Netflix or Prime Video follows suit, or if their scale makes this kind of integration unnecessary. For now, Tubi’s making a calculated play: trade some control over the user experience for access to 900 million potential viewers who are already in the habit of asking ChatGPT to solve their problems.