Text, Image, Video — What Comes After the Scroll?

A phone screen glows in a dark room. Someone has been scrolling for 38 minutes, eyes glazed, attention slipping. By morning, none of it will be remembered.

This is not a story about willpower. It’s a story about a content format running out.

A 2025 survey of social media users found that 64% report feeling “tired and exhausted” after sessions; 56% report burnout and trouble focusing; 55% feel more anxious than before they opened the app. Among Gen Z, 29% deleted at least one social platform last year. Even TikTok, in its own 2026 trend report, acknowledged it: the passive scrolling era is ending.

The interesting question isn’t what’s wrong with video. Video is fine. The interesting question is what comes next.

Four eras, not three

The internet has organized itself around content formats since the beginning. Each era was defined less by a technology than by a user mechanic — what people did when they were online.

The text era (1990s–2000s). Wikipedia, blogs, early forums and bulletin boards. The mechanic was reading and posting. Content was a thing you absorbed and occasionally replied to. The dominant unit was the article.

The image era (2000s–2010s). Instagram, Pinterest, Tumblr. The mechanic shifted to capturing and viewing. Phones got cameras, bandwidth got cheap, and the dominant unit became the photograph. Identity moved from what you wrote to what you saw.

The video era (2010s–2020s). YouTube, TikTok, Reels. The mechanic became filming and watching. Vertical video and the algorithmic feed turned content consumption into something closer to a slot machine. The dominant unit became the short vertical clip.

Each transition followed the same pattern: a new technology made the previous era’s mechanic feel slow, the unit of content got smaller, and engagement went up — until it didn’t.

We are now at the point where video engagement has stopped going up. Time on TikTok has plateaued in the US since late 2024. Reels growth has flattened. YouTube Shorts cannibalized long-form. Meanwhile, the cost of producing video has collapsed to near-zero, which means feeds are filling with AI-generated slop that users scroll past faster than ever.

The pattern history would predict is that something new arrives — not as an improvement on video, but as a different mechanic altogether.

The fourth era is interactive

The fourth era’s mechanic is describing and playing.

You don’t film something. You describe it, and an AI builds it. You don’t watch it — you touch it, tap it, shake your phone, talk to it. The unit of content isn’t a clip; it’s a tiny interactive moment. A playable.

This isn’t speculative. It’s already happening, distributed across products most industry observers haven’t connected yet.

Suno is an AI music generator: a text prompt produces a complete original song — vocals, lyrics, and instrumentation. Over 100 million people have used it. Most of them had never made music before.

Character.AI, founded by former Google LaMDA engineers, lets users create and chat with AI characters — fictional, historical, celebrities, anything. The average session runs around 17 minutes, well above ChatGPT’s 7.

These cover two corners of interactive content — AI-generated audio, AI-generated conversation. The third corner, and the most mainstream consumer expression of the pattern so far, is Loopit.

Loopit looks, on the surface, like TikTok: a vertical feed, swipe to advance. But every post is something you play with. A user describes an idea — “a game where falling grapefruits balance on a capybara’s head,” “an ASMR bubble-wrap toy,” “a meme of my cat dancing to lo-fi” — and Loopit generates a touchable mini-app that responds to taps, swipes, tilts, the microphone, the camera. Other users can remix it in one tap.

The numbers are what make Loopit interesting as a case study rather than a curiosity. The app launched in February 2026. By April, it had 1.44 million monthly active users, with 846,000 new first-time opens each month. iOS rating 4.85, Android 4.60 — both above what mature social apps usually hold. In April, Garena (a Sea Limited subsidiary) led a $50M round, bringing total funding to roughly $100M across three rounds in 30 days. Elon Musk reposted a user demo within days of launch. Chinese media has been calling it “the AI-era TikTok.”

The founder, Chen Weipeng, was previously head of R&D at Sogou Search and co-founder and LLM lead at Baichuan Intelligence. He frames the broader shift in one line: the past distributed information; the future distributes experience.

What’s actually changing

The first three eras all distributed information. A blog post, a photograph, a video — these are all packets of fixed information that move from a creator to a consumer. The user’s role is to absorb.

Interactive content distributes something else. It distributes the conditions for an experience. When you open a Loopit playable, the thing on your screen doesn’t exist until you touch it. Each user generates a slightly different session. The content isn’t a finished artifact; it’s a state machine waiting for input.

This collapses two things that have been separate for the entire history of the consumer internet:

Creation and consumption are merging. In the video era, 1% of users created and 99% consumed. The barrier was time, gear, editing skill. In the interactive era, “making” and “watching” take roughly the same amount of effort — you describe a thing, you get a thing. Loopit’s internal data already shows users who first opened the app to scroll converting to creators within their first session at rates that would be inconceivable for video platforms.

Content and software are merging. Every post in a Loopit feed is, technically, a small running program. Every Character.AI conversation is a live system. Every Suno track is a generated artifact. The distinction between “content” and “tool” — central to how the tech industry has been organized for twenty years — is starting to dissolve.

Both shifts have the same root cause: AI made describing as cheap as watching. Once that happened, the entire pyramid of who-makes-what-for-whom started restructuring itself.

Why most observers will miss this

The fourth era is easy to miss because it’s not arriving as a single product launch.

There is no “TikTok of interactive content” — instead there are several different products, each attacking a different vertical (music, conversation, playables) with the same underlying mechanic. To an industry trained on category leaders, this looks like noise. To someone watching the pattern, it looks like the early image era, when Instagram, Pinterest, and Tumblr each seemed like separate fads before anyone realized they were all instances of the same shift.

The second reason it’s easy to miss: most metrics in this era are still video metrics. Watch time. Completion rate. Shares. None of those are the right primitives for interactive content, where the right metric is something closer to “sessions started” or “remixes per playable.” Industry dashboards will keep showing video declining without showing what’s replacing it, because the replacement doesn’t fit in the same row.

And the third reason: a lot of the most interesting work is happening outside the US. Loopit is Chinese. The reverse-globalization pattern — where Chinese consumer AI products reach Western users before Western tech press notices — has been quietly building for two years and accelerated through 2026. By the time mainstream coverage catches up, the category will already be defined.

What to watch over the next 12 months

Three signals worth tracking.

The first is whether interactive content gets its own social graph. Right now, most of these products are feed-based but don’t have strong follow-graphs. The moment one of them does — the moment “follow this playable creator” becomes a thing the way “follow this TikTokker” did in 2019 — the category locks in.

The second is whether Apple and Google adapt their app stores. App Store taxonomy still doesn’t have a category for “interactive content platform.” Loopit is currently filed under Entertainment. When the stores get a proper category, that’s a signal the platforms have noticed.

The third is whether traditional social platforms try to copy this. Reels was Meta’s response to TikTok. The interactive-content equivalent — TikTok or Instagram launching a generative playable layer inside their existing feed — is the move I’d watch for in 2027. If they don’t make it, the category remains open. If they do, the four-era framework will become impossible to dismiss.

The internet has had three content eras so far. We’re already inside the fourth. The companies and observers who recognize the shift early will define the next ten years; the ones who keep measuring everything in watch time will spend the same decade wondering why their numbers stopped going up.

What comes after the scroll isn’t another scroll. It’s something you do, not something you watch.