❓ Problem

Two of the biggest behaviour shifts of the decade are happening in parallel, in separate apps, and nobody has bolted them together.

The first is short-form video, which has quietly become the default way Americans spend their downtime. The US has around 136 million monthly TikTok users and 82.2 million daily actives, with US adults averaging 52 minutes a day in the app. YouTube Shorts is doing 200 billion daily views. Advertisers followed the attention: short-form ad spend hit $111 billion in 2025, the fastest-growing category in digital advertising.

The second is prediction markets, which went from a niche crypto curiosity to a mainstream American pastime in about nine months. Combined monthly volume on Kalshi and Polymarket rose from under $5 billion in September 2025 to $44.8 billion in June 2026, a 75% jump on May alone. That is more than triple the $14 billion average monthly handleof every legal US sportsbook in 2025. Meanwhile 27% of Americans now hold an active online sports betting account.

Here is the kicker. Short-form video is enormous and monetises terribly for the people making it. TikTok's Creator Rewards pays roughly $0.50 to $1.00 per 1,000 qualified views, which means a creator needs a million views to buy a decent dinner. The whole format is optimised for the first 1.5 seconds, because views are the currency, which is why every feed is now bait thumbnails and manufactured cliffhangers that never pay off.

Predictions fix the wrong incentive. Views reward the hook. Predictions reward the payoff.

✅ Solution

Scrollcall. A short-form video app where every clip has a fork in it.

• The fork. Around the halfway mark the video pauses for three seconds and asks you to call it. Does he land the trick, does the recipe collapse, does she pick the red door. Two options, a thumb tap, no typing. Then it plays on and you find out within seconds.
• The stake is status, not money. You play with a points balance, a daily streak and a public accuracy rating. A profile becomes a track record instead of a follower count, and the people who can genuinely read a video climb a leaderboard.
• Creators are paid on prediction volume, not views. That single change rewires everything. The money moves from the thumbnail to the genuine unknown in the middle, so the incentive is to build content that is actually hard to call.
• Sponsors buy the prediction slot itself. It is the one moment in a short-form video where the user is provably not scrolling, thumb hovering, eyes on screen, waiting. That attention is worth several multiples of a standard in-feed impression, and unlike a view it is verifiable.
• The data exhaust is the moat. Every fork produces a crowd probability on an unscripted human outcome. Which clips split the room 50/50, which creators are best at being unpredictable, which users are consistently right. Nobody else has that dataset, and it makes the feed algorithm sharper than a watch-time model ever gets.

The wedge is deliberately small: a daily game of ten clips and ten calls, Wordle-shaped, shareable, cheap to build. The infinite feed comes after.

📊 Key Numbers

TAM

• US short-form video is the attention layer: 82.2 million daily TikTok users in America before you count Reels and Shorts, feeding into a $111 billion global short-form ad pool, of which the US is the single largest slice.
• The free-to-play comparable is social casino, worth roughly $2.8 billion a year in the US on virtual currency alone, with no real money changing hands.