Cardboard Wants to Be the Cursor for Video, and That Comparison Is Doing a Lot of Heavy Lifting

The Macro: Everyone Is Trying to Fix the Part of Video That Isn’t Shooting

Here’s the thing about video content in 2025: the bottleneck was never the camera. Phones are absurdly good. Capture is solved. What kills people is the hour (or four) that comes after, staring at a timeline, dragging clips around, trying to remember which take had the good audio.

Marketing teams know this acutely. Video consumption keeps climbing and budgets have to stretch further than they used to, which means the demand for fast, competent editing has never been higher without a corresponding spike in editors available to do it. The digital marketing space is growing fast, multiple sources peg it well above 10% CAGR, and a huge chunk of that is video-first content.

So the tools market responded. You’ve got Descript, which came at it from the transcript angle. CapCut, which came at it from TikTok virality and sheer mobile ubiquity. Adobe Premiere trying to bolt AI onto a 30-year-old interface. Runway doing genuinely weird generative stuff. And then a bunch of smaller players doing one specific thing well, like auto-captions or B-roll matching.

What nobody has really cracked is the agentic layer. The thing that doesn’t just assist you but actually drives. Cursor cracked this for code. You describe intent, the agent executes, you review and redirect. It turns out that model works well when the underlying medium is structured enough for an AI to navigate. Code is structured. Video, it turns out, kind of is too. Clips have metadata, duration, audio waveforms, faces, scene cuts. There’s a lot for a model to grab onto.

That’s the gap Cardboard is trying to fill. Whether the analogy fully holds is the interesting question.

The Micro: The Browser Is Now the Editing Bay

Cardboard is a browser-based, agentic video editor. The whole pitch is on the homepage: your browser is about to become a studio. You upload raw footage, tell it what you want, and it figures out the edit.

According to the product description, it understands what’s in your clips, has “the taste to know what a good edit looks like,” and executes your vision. That last phrase is doing real work. Taste is the hardest thing to encode, and claiming it is a bold swing.

The Cursor comparison, which they made themselves in the tagline, is useful framing even if it sets a high bar. Cursor works because it doesn’t just autocomplete, it holds context, reasons about structure, and takes initiative without completely going rogue. If Cardboard has actually built something close to that for footage, that’s genuinely interesting. If it’s closer to “smart auto-edit with a chat box,” that’s a different product with a better tagline than it deserves.

The founding team has some credibility here. Stein Magnus Jodal, co-founder and CEO, previously worked as an engineering manager at Oda, according to Crunchbase. Ishan Sharma is co-founder and CTO, according to his LinkedIn. Saksham Aggarwal is also listed as co-founder. Notably, Cardboard is a YC W26 company, which means it went through the batch with that cohort’s level of scrutiny and survived. That’s a real signal.

It launched and got solid traction, landing near the top of the charts on the day it went live.

The product is currently invite-gated, sign up at usecardboard.com. No pricing is visible, which either means it’s free during beta or they haven’t figured that out yet (both are plausible for a YC company at this stage).

I’d love to know how it handles multicam, or footage with messy audio, or the very human problem of “I know a good moment when I see it but I can’t describe it.” Those edge cases are where agentic editors either earn the label or quietly reveal they’re just a smarter ffmpeg wrapper.

For context on what AI-native creative tools can actually pull off, what Picsart is building with Persona and Storyline shows how far the “AI as creative collaborator” thesis can go when a team with real infrastructure commits to it. And if you want a case study in what happens when a solo founder uses AI tools to sidestep an entire industry’s cost structure, the Lunair story is worth reading before you dismiss the solo-operator use case for something like Cardboard.

The Verdict

I think the concept is right. The timing is probably right too. The gap between “I have footage” and “I have an edit” is genuinely painful, and the Cursor framing, borrowed as it is, actually helps people understand what agentic means in this context without needing a six-paragraph explainer.

What I don’t know yet: whether the taste claim is real. That’s where this lives or dies. If Cardboard’s model genuinely understands pacing, story arc, what to cut for rhythm versus what to cut for time, then it’s a serious product. If it produces technically competent but emotionally flat edits that a human still has to redo from scratch, then it’s a time-saver at best and a source of frustration at worst.

At 30 days I’d want to know what the retention looks like on early beta users. At 60 days, whether any marketing teams have replaced a real workflow with it. At 90 days, how they’re handling the pricing conversation once the free beta ends.

YC backing means the next six months have some runway behind them. The question is whether the product earns the comparison it’s been given.