The yapper: AI gossip-format ads at 20 minutes per video
Girl in bed. Messy bun. Claw clip. She tells you her boyfriend was cheating. Plot twist. The girl is AI. He's running a course. About to buy her a ring. The audience never sees the pitch coming.
Storytime drama, but the talent is rendered, the drama is scripted, and the course plug at the end is the entire point. Under 20 minutes per video. A $2K/video UGC creator gets replaced by a 4-agent stack: script writer for the tension-twist-product structure, persona render for the in-bed talking-head, caption timer for the swipe-block beats, and a swap agent that rotates the product plug every cycle. People stay for the drama. They absorb the ad without realising it happened.
Write the gossip script: tension, twist, product
The script agent writes the storytime as a relatable confession. Sets up emotional tension. Builds curiosity. Lands the twist where the product is the punchline. The viewer thinks they're hearing a friend's drama. They're hearing a sales script structured like one.
The format pre-dates AI by a decade. Storytime TikTok has worked since 2019. The agent just industrialises it. Tension first, twist second, product third. The brand isn't writing an ad. The brand is writing a story that happens to include their product as the resolution.
Apply this: Stop writing ads. Write 60-second stories where your product is the plot twist. The pitch lands when the story resolves.
Render the in-bed talking head: FaceTime energy, not studio energy
The persona agent renders the talking head doing casual in-bed storytime. Messy hair, natural gestures, eye contact that feels like FaceTiming a friend. The lighting is iPhone-screen-lit, not ring-light-lit. The framing is tilted up from the pillow, not aimed straight from a tripod.
The framing IS the trust. A perfectly-lit shot reads as ad. A tilted-up-from-the-pillow shot reads as content. Same persona, two different framings, two different conversion mechanisms. The brands that get this right ship gossip storytime ads at the framing fidelity of a Sunday-morning vlog.
Apply this: Render the camera, not just the persona. The amateur framing is what kills the ad-filter in the viewer's head.
Add pink keyword captions timed to the emotional beats
The caption agent adds pink keyword-highlighted captions timed to the emotional beats. "Glued to his laptop". Pink. "Takes it to the bathroom". Pink. The captions hold attention through the parts where audio-off scroll would otherwise kill the watch.
80% of feed scrolls happen with audio off. The brand that ignores that ships the gossip script as voice-over and loses 80% of the audience. The agent shapes the captions around the beats the algorithm rewards. The moments where curiosity peaks and the swipe-block becomes essential.
Apply this: Caption every emotional beat. Pink, oversized, on-screen, timed to the line. Audio-off viewers are the majority. Design for them first.
Swap the product, keep the format: one engine, every niche
The swap agent tests new story angles and rotates the product plug for the next niche every cycle. Same format, different product, every 20 minutes. The script structure is the constant. The brand mention is the variable.
This is the scaling unlock. A brand running one product can use this engine for one niche. A brand running 10 products can use the same engine across 10 niches with one stack of agents and one script template. The unit cost per ad collapses as the product portfolio expands.
Apply this: Build the engine once. Swap the product line. Don't rebuild the engine for every launch. That's where most operators waste their budget.
Storytime that happens to sell: the ad-filter never fires
The viewer stays for the drama. They absorb the ad without realising it happened. The conversion mechanism isn't persuasion. It's entry through the side door. By the time the product plug lands, the viewer has already invested 50 seconds of emotional attention. The pitch arrives wearing a story.
Direct-response ads announce themselves. Gossip storytime ads never do. The brand that runs this format isn't trying to convince. They're trying to entertain. And let the conversion happen as a byproduct of attention well-held.
Apply this: Stop running ads. Start telling stories that happen to sell things. The conversion is the side effect, not the goal.
- "I replaced $[X]/video UGC creators with AI [format] scripts"
- "Under [N] minutes per video. The audience never sees the pitch coming"
- "Plot twist. The [character] is AI. [Brand] is running a [course/product/launch]"
- "People stay for the [drama/story/twist]. They absorb the ad without realising it happened"
- "It's not about running ads anymore. It's about telling stories that happen to sell things"
What’s actually running underneath
- Script agent (Claude tension-twist-product) Writes the gossip storytime as a relatable confession. Sets up emotional tension, builds curiosity, lands the twist where the product is the punchline. Same Claude agents we run across the network. The structure pre-dates AI, the writing speed is what changed.
- Persona agent (Seedance 2.0, in-bed framing) Renders the talking-head persona doing casual in-bed storytime. Messy hair, natural gestures, FaceTime-energy eye contact. iPhone-screen lighting, tilted-up-from-pillow framing. Not studio lighting. The framing is what kills the ad-filter.
- Caption agent Adds pink keyword-highlighted captions timed to the emotional beats. Captions hold the audio-off viewer through the parts where they'd otherwise swipe. 80% of feed views happen audio-off. This is where the format earns its retention.
- Swap agent Rotates the product plug for the next niche every cycle. Same format, different product, every 20 minutes. The script structure is the constant; the brand mention is the variable. One stack, every product line.
A standard UGC creator deal runs $2,000 per video. This stack ships gossip storytime ads in under 20 minutes per video at cents per render. The unit economics flip from agency-grade to render-grade.
The brands still paying $2K per UGC creator read are buying performance-by-luck. The brands running this stack are telling stories that happen to sell things. At the unit cost of the render, not the talent. That gap compounds across a year of product launches.
Want this shipped on your product in 2-3 days?
Book a 15-min call. I’ll show you exactly how the pipeline adapts.
Book a call →