Netflix revamps its TV app and tests generative AI search on iOS

Netflix is tired of watching you spend 45 minutes browsing thumbnails only to end up rewatching The Office.

Netflix Chief Product Officer Eunice Kim and Chief Technology Officer Elizabeth Stone pulled back the curtain on a major overhaul of the Netflix TV homepage, alongside new mobile discovery experiments powered by generative AI.

A streamlined TV app and TikTok-style discovery

The most noticeable change on the big screen is layout ergonomics. Netflix moved the Search and My List shortcuts from the semi-hidden left rail straight to the top of the screen, making navigation far more intuitive.

Title cards now put decision-making context front and center. Instead of making you dig through descriptions, the interface highlights key hooks like Emmy Award Winner or #1 in TV Shows right away. Under the hood, recommendation algorithms have been tuned to be more responsive to your real-time moods.

Over on mobile, things get even more experimental. Netflix is rolling out a small opt-in beta on iOS that introduces generative AI to search, letting members find movies with conversational prompts like "I want something funny and upbeat."

Netflix will also begin testing a vertical video feed packed with short clips from its catalog. If a scene grabs your attention, you can tap directly to start the full movie, add it to your list, or share it with friends.

Why it matters

For viewers, this update targets the single biggest pain point in modern streaming: decision fatigue. If natural-language conversational search works as advertised, it finally solves the problem of knowing the vibe you want to watch without knowing a title or actor name.

For the streaming industry, the battleground has shifted from pure catalog size to discovery efficiency. Platforms are no longer just competing against each other; they are fighting TikTok, YouTube, and short-form video for evening screen time.

We think deploying generative AI for vibes-based search is an exceptionally smart move. Traditional metadata tags struggle with nuanced human mood queries, which is precisely where conversational models shine.

That said, leaning heavily into a vertical snackable feed carries risk. It might hook users instantly, but it also trains audiences to expect instant dopamine, potentially making it harder to convince them to commit to a slow-burn prestige drama.