Most marketing teams approach video editing the way you might approach a dental procedure: grimacing, postponing it until absolutely necessary, and—if budget allows—paying someone else to handle it. The software isn't the problem. Premiere Pro has been around since 1991. DaVinci Resolve is free and staggeringly powerful. Final Cut has its devotees.
But these tools were designed for people who know what they're doing. Professional editors. The kind who understand the difference between J-cuts and L-cuts, who can navigate a twenty-layer timeline without breaking a sweat. They weren't built for the growth marketer who needs three video ads by Friday, or the founder who wants to stitch together customer testimonials without spending $800 on a freelancer.
Enter Cardboard.
The company—just two people, both engineers, currently grinding through Y Combinator's Winter 2026 batch—thinks the solution isn't yet another video tool. It's teaching the computer to think like an editor. A junior one, maybe, but an editor nonetheless: someone who understands what you're trying to accomplish and handles most of the grunt work without you needing to explain how keyframes work.
On February 26, Cardboard surfaced on Hacker News. The post drew 129 upvotes and 80 comments in 48 hours. Founders Saksham Aggarwal and Sharma Ishan waded into the discussion, fielding questions about rendering pipelines, text-kerning algorithms, and whether browser-based video editing could ever match desktop apps on quality. The conversation stayed technical—curious rather than dismissive—which is about as good an outcome as a startup can hope for in that corner of the internet.
Drop Your Footage, Describe What You Want
The pitch is almost disarmingly simple. You drag raw video files into a browser window. Then you tell Cardboard what you want in plain English. "Find the moment where she talks about pricing and put it after the intro." Or: "Make this feel more energetic."
The system—using what Aggarwal and Sharma describe as "a series of Cloud VLMs plus traditional ML models"—analyzes the footage. It figures out what's being said, when faces appear, where the pacing sags. Then it generates a rough cut across a multi-track timeline.
All of this happens in your browser. No plugins. No uploading gigabytes to a server farm and waiting for renders to crawl back. The entire editing engine runs client-side, powered by WebCodecs and WebGL2. Since November, the founders have pushed 13 releases, adding features that sound like they belong in professional software: multi-track timelines, keyframe animation, shot detection, beat-sync through percussion analysis, voiceover generation and cloning, background removal, multilingual captions with spatial awareness.
You can export to Premiere Pro, DaVinci, or Final Cut via XML if you decide you want a human editor to take over later.
When Hacker News commenters questioned whether browser-based rendering could match desktop applications on quality and deterministic operations, Sharma dove into the thread with specifics. Harfbuzz compiled to WebAssembly for text rendering, for instance. The founders acknowledged the tradeoffs. They seem less interested in replacing professional editors than in collapsing the time between "we need a video" and "here's a first cut worth refining."
Which, honestly, might be the smarter play.
The Pedigree
Aggarwal studied computer science at BITS Pilani, built AI products at Iterate AI (an Entrepreneur First-backed venture), published research at ACL, and joined PYOR—a startup backed by Coinbase and Castle Island—as its first engineer. Before that, he worked on growth at Flint, which had Sequoia money behind it.
Sharma spent roughly five years at HackerRank, itself a YC alum from the Summer 2011 batch. He built performance-critical browser applications there and led multiple zero-to-one products. His background is in high-performance web systems and creative tooling—the kind of deeply technical work that makes real-time video processing in a browser sound almost routine.
They're calling Cardboard an "agentic" editor. The term is having a moment across AI tooling, perhaps more than the founders expected. Adobe announced creative agents and a "Quick Cut" beta for Premiere just weeks ago. YC's Winter 2025 batch included Mosaic, another company building agentic video workflows, this time with node-based multimodal agents. Runway raised $308 million last year.
The narrative is converging: video editing is the next creative workflow to get the ChatGPT treatment, where describing the change you want beats making it manually.
Early Traction, Ambitious Roadmap
Cardboard's homepage displays logos from early customers: PostHog, Autumn, General Legal, Hyperspell, Oolka, Oximy, Shopos. Pricing starts at $60 per month—not cheap, but not outrageous if it actually saves your team a few hours each week.
The company launched with a no-login demo, a choice that paid off on Hacker News. People could try the product immediately, which kept the conversation technical and curious rather than skeptical and dismissive.
The roadmap is ambitious for a two-person team three months into YC. Real-time collaboration is coming—Aggarwal and Sharma describe it as "video git"—along with a prediction engine that learns your editing patterns and suggests next moves before you ask. Y Combinator's standard deal gives them $500,000 ($125,000 for 7 percent equity on a post-money SAFE, plus $375,000 on an uncapped most-favored-nation SAFE). Demo Day is March 24, 2026.
Whether they can execute on all of this remains to be seen. But they've already shipped 13 releases since November, which suggests they understand how to move fast.
The Bet They're Making
Cardboard isn't trying to replace Premiere Pro any more than Notion tried to replace Microsoft Word. Different job entirely. It's aiming at the moment between "we have raw footage" and "we have something good enough to show the team"—the stretch where most marketing videos currently die in Google Drive folders because no one wants to spend four hours learning how to sync audio tracks.
The fact that it runs entirely in the browser—and that Aggarwal and Sharma are publicly wrestling with WebCodecs limitations on Hacker News—suggests they're betting on the web as the platform for the next generation of creative tools. Not a guaranteed bet, but not a crazy one either.
Whether it pays off depends on how well the agentic editing actually works when real teams use it under deadline pressure, and whether $60 a month feels like a bargain or just another SaaS subscription bleeding your budget.
But the early traction, the velocity of shipping, and the technical depth both founders bring to the problem suggest they've identified something real: the bottleneck in video marketing usually isn't the final polish. It's getting from zero to one fast enough that the footage is still relevant when you finally publish.
That's the problem Cardboard is solving. We'll find out soon enough if the market agrees.
