Yarn Wants Video to Feel Like Writing a Doc
The Y Combinator W24 startup lets go-to-market teams make product and sales videos from a script - no editor, no agency, no month of back-and-forth. Founders Jasper Story and Nicole Atack are betting the bottleneck was never ideas. It was production.
Every founder knows the drill. You want a product video. You write a rough script, hunt for an editor or an agency, wait two weeks, get something that looks nothing like your brand, send notes, wait again, and by the time it ships the launch is already old news. The video was never the hard part. Everything around the video was.
Yarn, a company from Y Combinator's Winter 2024 batch, is built on that specific frustration. Its pitch is short: make sales and marketing videos as easily as you write a document. You type a script, and the software handles the voiceover, the screen recording, the on-brand overlays, and the edit. The company says a founder can go from blank page to finished marketing video in under 20 minutes - and, by its own math, at roughly 95% less than the cost of an agency.
That is a big claim in a crowded field. But the interesting thing about Yarn is not the speed number. It is the shape of the bet underneath it: that the reason most startups barely make video is not laziness or lack of ideas. It is that the workflow is genuinely miserable, and no one has made it feel native to how small teams actually work.
01 / THE IDEAVideo, but you edit the words
The core interface is the tell. In most video tools, the timeline is king - a strip of clips you drag, trim, and nudge frame by frame. Yarn leads with text. You write what you want said, and the video assembles around the script. Change a sentence and the video changes with it. Need a different voice? Pick one from a library, or match your own. Want to reuse the same demo in three formats? The script travels; the edit adapts.
That inversion - script first, timeline second - is what makes the "like a doc" comparison more than marketing. It moves video out of the hands of specialists and into the hands of the person who actually knows what the product does. The founder writes. The tool produces.
One of Yarn's more unusual pieces is what it calls a part-synthetic talking head: record yourself once, then generate new sentences from that footage so you are not re-shooting the same intro for the fiftieth time. It is the kind of feature that sounds like science fiction and lands as plumbing - a way to kill the most repetitive task in go-to-market video without tipping into the uncanny.
02 / HOW IT WORKSFour steps, no editor
The output is meant to look designed, not dashed off. Yarn offers brand control down to motion blur, cursor effects, and zoom transitions, plus generated overlays you can also import from Figma or Google Slides. The founders have said, plainly, that they want the output to sit next to tools like Notion and Linear on the quality scale - polished by default rather than polished if you fight for it.
03 / THE FOUNDERSAn ad-maker and a physicist
The pairing behind Yarn is not the usual two-engineers story. Jasper Story, the CEO, spent time making ads for Google and worked as a designer and product manager at early-stage startups - and, in an earlier life, taught high school math. Nicole Atack, the CTO, came to startups from quantitative finance and particle physics. One founder knows what a good ad looks like and why people watch it; the other knows how to build the models that make the software feel seamless.
It is a fitting division of labor for a company whose whole promise is that taste and automation can live in the same product. And yes - the CEO of a video storytelling company is named Story. Some things you cannot make up.
On LinkedIn, video is shared far more than any other post type - a stat Yarn leans on in its pitch. If video is the format that travels, the company that makes video cheap to produce is selling distribution, not just editing.
04 / THE MARKETNot fighting Synthesia head-on
The AI video space is a knife fight. Synthesia and HeyGen own the corporate-avatar corner. Descript reinvented editing around a transcript. Loom made the quick screen recording a default. Arcade and Tella chase interactive product demos. Into that, a small company could easily get flattened by trying to do everything.
Yarn's answer is to narrow. It has repositioned as "the all-in-one for GTM video" - go-to-market video for startup teams, specifically. Not filmmakers, not YouTubers, not enterprise training departments first. The buyer is a founder, a head of sales, or a scrappy marketing lead who needs launch videos, demos, and sales collateral this week. Narrow is underrated: it is easier to be the obvious choice for one crowd than a passable option for everyone.
The distinction from Loom is the cleanest way to understand the wedge. Loom captures raw screen. Yarn's ambition is a finished asset - edited, branded, voiced, and ready to send to a Fortune 500 prospect. The comparison Yarn invites is not "faster screen recorder" but "in-house creative studio you can afford."
05 / CUSTOMERS & MODELWho's paying, and how
The early customer list skews toward Yarn's own world: Clay, plus YC peers like DriverAI, Replo, and Forge, using it for sales demos, marketing content, and procurement collateral. DriverAI has used it to spin up per-role sales demos for large prospects - the kind of personalization that used to require a bespoke shoot for every buyer.
The business is B2B SaaS on a credit model. The entry Growth tier runs about $1,240 a month for a single seat and a bundle of annual credits, with Scale and Enterprise tiers priced on request and stacked with more credits, seats, and templates. This is not a hobbyist tool anymore; the pricing makes clear Yarn is aiming at teams that would otherwise be cutting agency checks.
06 / THE STAKESWhat has to be true
Yarn raised a seed round backed by Y Combinator, General Catalyst, Leonis Capital, and angels from Pika Labs, Roblox, and Adobe. That is a capable cap table for a company this young, and it buys time - but the hard question is durability. Yarn does not train the frontier video models it uses; it orchestrates them, wiring Sora, Veo, Kling, and Nano into one workflow. The moat, if there is one, lives in the workflow, the taste baked into the defaults, and the retention its personalized videos can actually deliver.
That last number is the one to watch. Yarn cites 90.1% viewer retention on AI-personalized content through its Content API. If that holds up outside a demo, it reframes the whole product: not a cheaper way to make video, but a measurably better-performing one. Speed gets you a video. Retention gets you a customer. Whether Yarn can keep proving the second point - as the models it rides keep shifting under it - is the story worth following.
For now, the bet is legible and the wedge is sharp. A small team is trying to make the thing every startup says it should do, and rarely does, boring enough to actually happen. If they are right that production was the whole bottleneck, a lot more startup video is about to get made.
Explore Yarn
Watch & read
Sources: yarn.so, Y Combinator, LinkedIn, and public press. Figures marked approximate are company-stated or estimated and may change.