EN
Back to the archive

The archive · Developer & Business Tools · Product decision · 2025–2026

Cardboard bets AI first cuts replace manual editing; PH #2 of day, YC W26

Two friends who hit the same editing wall build a browser agentic video editor that turns raw footage plus a goal into a first cut in minutes; YC W26.

Cardboard

The betVideo editing will move from manual timelines to agentic first cuts: describe the goal and an AI director proposes an editable draft, making iteration speed the new moat.Live

What the business is

Cardboard is a browser-based agentic video editor that takes raw footage plus a natural-language goal and produces a first cut on a multi-track timeline the user can refine.

How it started

Saksham Aggarwal (ex-CTO of EF-backed Iterate AI, ~250K views/month on socials) and Ishan Sharma (five-plus years of performance-critical browser work at HackerRank) kept hitting the same wall: creating content was fast, editing was slow. In late 2025 they built Cardboard and had shipped 13 releases by the time YC's W26 batch ran.

What happened

Cardboard launched on Hacker News on 2026-02-26 as 'the Cursor for video editing', drawing 132 points and 83 comments; on 2026-03-11 it hit #2 of the day on Product Hunt with ~450 upvotes. YC lists it as active with a five-person team in San Francisco/Bengaluru; the product keeps adding beat-sync, voice cloning, and Premiere/DaVinci XML export.

How it ended up

Still live as of September 2026 — YC lists Cardboard as an active W26 company with open jobs; no funding round or revenue figures have been published.

Background

Cardboard is a browser-based agentic video editor. Users give it raw footage plus a goal — 'make a 60s recap', 'cut this into a 20s ad' — and it proposes a first cut on a multi-track timeline that the user can then refine. Everything renders client-side on WebCodecs/WebGL2, with video understanding handled by cloud VLMs, so there is nothing to install and no server-side render pipeline.

The bet was that video editing would move from manual timelines to agentic first cuts: instead of competing with Premiere or DaVinci on fidelity, Cardboard competes on speed, letting an 'AI director' make creative decisions while humans stay in command. The founders lived the problem themselves — one creates content at ~250K views/month, the other produced launch videos for HackerRank — and met it from opposite directions.

Traction came fast. The Launch HN on 2026-02-26 drew 132 points and 83 comments, and on 2026-03-11 Cardboard ranked #2 of the day on Product Hunt with roughly 450 upvotes. As of September 2026 YC lists the company as active with a five-person team; the product keeps shipping features like beat-sync, voice cloning, and XML export to Premiere/DaVinci, but no funding round or revenue has been announced.

What has to be true

  • Both founders hit the same wall independently: creating content was fast, editing was slow, so the product was a solution to their own workflow.
  • Client-side rendering removed install friction, letting HN and PH users try the real editor, not a demo.
  • Iteration speed became the pitch: get a strong first cut in minutes and refine, instead of hours of scrubbing footage.
  • The YC W26 stamp added credibility while the team stayed small and product-focused.

What can be applied

When incumbent tools are powerful but painfully manual, the wedge is speed to first cut, not fidelity: let the agent produce the draft and humans refine, and HN/PH do the validation.

Aftermath

As of 2026-09-02 Cardboard is active: YC's company page lists a five-person team, open jobs in Bengaluru and San Francisco, and an active status. The product page remains live and the company continues to frame itself as the future default workspace for video production, the way Canva/Figma became defaults for design. No further funding or revenue figures have been published.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases