Netflix starts testing AI Discover on TV to cut through the scroll

Netflix began rolling out an experimental AI Discover tab on U.S. TVs on Nov. 3, using mood- and theme-based recommendations to help viewers find something to watch more quickly. The feature is part of the company’s latest effort to make TV browsing less tedious and more responsive to what people say they want in the moment.

What Netflix launched

The new tab appears in Netflix’s TV navigation as Discover, marked with a red gem icon and a New label, according to TechCrunch. Netflix is positioning it as an experimental discovery surface rather than a full replacement for the current home screen.

The first version focuses on mood-based browsing. Instead of asking viewers to type a title or scroll endlessly through rows, AI Discover offers prompts such as something light, scare me, family movie night, and quick watches. The idea is simple: let users state a mood, then let Netflix do the sorting.

Netflix says the system uses a new intelligence layer that learns from real-time user input. In practice, that means the recommendations are meant to adapt as people refine what they want. A viewer can move from broad to narrower themes, or restart with something different if the first pass misses the mark.

How does AI Discover work?

How does Netflix’s AI Discover tab work? It works by translating simple mood or theme prompts into recommendation sets that update as users give feedback.

On title pages, Netflix is also adding a feature labeled Tune Your Recommendations beside Play and My List. TechCrunch reported that this tool lets users save titles they like and exclude ones they do not want to see again, adding another layer of input to the recommendation engine.

The company is framing the tab as a place to experiment. Netflix described it as a

test kitchen

for discovery ideas, suggesting the company is treating the rollout as an early trial rather than a finished product. The phrase matters because Netflix is not just changing a menu label; it is testing whether AI can make the company’s biggest interface problem easier to solve on a television set.

Why it matters

Netflix lives and dies by discovery. With a catalog that is always larger than what any one person can browse in a few minutes, recommendation quality directly affects whether a viewer keeps scrolling, starts something new, or quits in frustration.

That problem is especially sharp on TV. Typing on a remote is slower than tapping on a phone, and even a strong recommendation engine can feel blunt when a viewer is in a specific mood. Netflix’s AI Discover tab is an attempt to narrow the gap between how people think about entertainment and how streaming interfaces typically work.

The launch also arrives at a moment when Netflix is under pressure to keep attention high. The Verge reported that the company recently laid off around 1,000 employees amid slower growth and weaker viewing. In that context, a more efficient discovery system is not just a product tweak. It is part of the company’s broader effort to keep engagement strong and reduce friction inside the service.

There is another business reason to watch this closely: if the feature works, Netflix can learn which kinds of prompts, moods, and viewing intents actually convert into play starts. Those signals could shape not only the TV app, but also how Netflix thinks about surfacing content across its wider product.

What makes this different from normal Netflix search?

What makes AI Discover different from ordinary Netflix search? It is designed to start from the viewer’s intent, not from a title, actor, or genre name.

Traditional search is useful when a user knows what they want. The new tab is for the more common streaming scenario: a person knows the feeling they want, but not the title that fits it. That is why the initial themes lean into moods and situations rather than categories alone.

This also helps explain why Netflix is testing the feature on TV first. On a living-room screen, browsing can be awkward and slow, while a simple prompt-based flow may feel more natural than navigating rows with a remote. If the company can reduce the effort required to choose something, it may increase the odds that people finish the decision and hit play.

What is in the first version?

The first rollout is limited, and the details matter. According to TechCrunch, the feature is starting in the United States for English-language users on supported TVs. What’s on Netflix said the initial rollout is going to a small group of U.S. members today, which suggests the test is still narrow and controlled.

There are also important limits on who can use it and how. What’s on Netflix reported that AI Discover is restricted to standard profiles and does not appear on kids profiles. The site also said there is no chat box in the first version, which means Netflix is not launching a conversational assistant here, at least not yet.

Netflix has also not said whether AI Discover will expand to mobile, other languages, or other countries, according to What’s on Netflix. The company will publish a Help Center article with compatible-TV details, suggesting device support may be selective at first.

By the numbers

Background: how Netflix got here

Netflix has spent years refining recommendations because discovery is central to streaming. Unlike a linear TV channel schedule, a streaming service asks viewers to make a choice from a giant catalog every time they sit down. Recommendation systems are the machinery that keep that experience manageable.

That challenge is even more obvious on television than on mobile or web. Browsing with a remote is slower, text entry is less convenient, and the choice overload can become part of the experience. Netflix has already built a reputation around personalized rows and tailored suggestions, but the company appears to be pushing that logic one step further by making the interface more explicitly conversational in style, even without an actual chat window.

What’s on Netflix also noted that Netflix tested an opt-in generative AI search beta on iOS in May 2025. That earlier experiment suggests the company has been probing AI-assisted discovery before bringing a version of it to the TV app. The TV rollout therefore looks less like a sudden pivot and more like a staged extension of an existing idea.

At the same time, the sources leave some room for interpretation. TechCrunch’s reporting emphasizes the new intelligence layer and the experimental “test kitchen” framing. The Verge highlights the gradual nature of the rollout to English-speaking U.S. members on supported TVs. What’s on Netflix, meanwhile, adds practical limitations such as the small initial audience, the lack of a chat box, and the absence of any public word on international expansion. Together, they paint a picture of a cautious launch, not a wide release.

What viewers can actually do

In the current version, viewers can choose broader or narrower themes to shape the results. If the initial suggestion set is off, choosing something different restarts the search. That may sound like a small interaction design detail, but it is a core part of how the feature works: the system is designed to be corrected quickly, not just observed passively.

Netflix is also trying to make the recommendation loop more explicit. The “Tune Your Recommendations” option gives users a way to mark what they like and what they do not want, feeding the system with more signals than a simple thumbs-up or thumbs-down model might capture.

The result is a discovery flow that aims to feel less like search and more like guided browsing. Instead of hunting through genres and rows, the viewer starts with an intent such as a light watch or a family night selection, then narrows from there.

Why TV is the real test

TV may be the hardest place to make this work, which is exactly why it matters. On a laptop or phone, a viewer can search directly and adjust quickly. On TV, even small interface improvements can save time and reduce frustration because the device itself is built for passive viewing, not active exploration.

If AI Discover reduces the number of steps between “I want something easy” and “play,” that is a meaningful product gain. If it does not, the feature risks becoming another button in an already crowded interface. The value of the experiment will be measured less by the novelty of the AI label and more by whether it changes behavior.

That makes the broader rollout logic important. Netflix is starting with supported TVs, English-language users in the U.S., and a small initial group, according to the reporting. That is a classic test-and-learn posture: limit the audience, watch how people respond, and decide whether the tool is worth scaling.

What happens next?

The biggest unanswered question is how wide Netflix plans to go. What’s on Netflix said the company has not disclosed whether AI Discover will expand to mobile, other countries, or other languages. That means the immediate future of the feature is still open, and the first public test may be used to decide whether it deserves a broader life.

Several other questions remain open as well:

• How many users are included in the first test group?
• Which TV models are compatible?
• Will the tab gain a real conversational interface later?
• Will Netflix add more prompt types beyond the initial mood themes?
• Will the learning from AI Discover affect the main Netflix TV experience?

Watch for Netflix’s Help Center details on compatible TVs and any follow-up announcements about expansion. If the company sees strong engagement, this could become a template for a broader AI-assisted discovery layer across the service. If the response is lukewarm, it may remain a limited experiment inside Netflix’s “test kitchen.”

How the outlets line up

TechCrunch focused on the product mechanics: the Discover tab, the mood-based prompts, the new intelligence layer, and the “Tune Your Recommendations” tool. The Verge added the broader company context, including the slower growth and weaker viewing that have followed Netflix’s layoffs. What’s on Netflix supplied the most granular rollout details, including the small initial audience, the kids-profile restriction, and the absence of a chat box.

Those differences do not conflict so much as they fill in different parts of the same picture. The common thread is clear: Netflix is trying a more intuitive, mood-driven way to help people find something to watch on TV, and it is doing so carefully, in a limited test, with no promise yet that the feature will travel beyond the U.S. TV app.

For now, the company’s bet is straightforward. If viewers can tell Netflix how they feel instead of what title they want, the service may be able to turn a tedious scrolling session into a faster decision. Whether that turns into a lasting change in how people browse streaming TV is the question Netflix is now trying to answer.