About

One person builds this

FirstCut is designed, built, and run by Eran Strassburg, a solopreneur. There is no team behind the “we” in these pages — when something works, that's me, and when something breaks, that's me too.

Why it exists

Anyone who has cut an interview knows the shape of the problem. The footage is three hours long. The story inside it is eleven minutes. And between those two facts sits a day of scrubbing a timeline, reading a transcript in another window, and writing down timecodes on paper — the part of editing that is pure clerical work, done before the actual editing can start.

That first pass is a reading problem, not a picture problem. A machine that reads a transcript well can do most of it: find where the subject actually answers the question, drop the throat-clearing, and put what's left in an order that holds together. That's what FirstCut does. It hands you a rough cut and a set of real timecodes, so you start the day where you used to arrive at lunchtime.

The other half of the idea is what it doesn't do: your video never uploads. Only the small audio track is extracted in your browser and sent for transcription; the multi-gigabyte files stay on your disk, and the preview plays from them directly. That's a deliberate constraint, not an optimisation — it's the reason you can put an unreleased edit through this at all.

What it is, and what it isn't

It's honest about being a first draft. The cut is a machine-made starting point that you take into your real editor — it will make choices you disagree with, and it's built to be argued with rather than accepted.

• It's a rough cut, not a final cut. The output is a structure to react to. You reorder it, mark what to keep and what to avoid, and hand it back to be refined.
• The player is a preview, not an NLE. Clip boundaries land within a few hundred milliseconds. The exported EDL, FCPXML or OTIO carries the exact timecodes — that's the deliverable, and it's what your editor should trust.
• It reads, it doesn't watch. Decisions come from what was said. It has no opinion on whether a shot is in focus or whether someone blinked.
• It selects; it doesn't generate. FirstCut is an independent product built on third-party speech-to-text and AI language-model providers, and isn't affiliated with or endorsed by them. The model picks and orders moments from your footage; it never creates images, video or audio. What it may be used for is in the Acceptable Use Policy.
• It's not storage. FirstCut never holds your video, which also means it can never give it back to you. Keep your own files and your own backups.

How it's run, and how to reach me

Support is one inbox — support@firstcut.video — read by the person who wrote the code. That means you get an answer from someone who can actually fix the thing, and it means I'm not going to promise you a one-hour response time. Most mail gets a reply within a couple of working days.

The same goes for the roadmap: bug reports and feature requests go straight to the person who decides what gets built next. If something in FirstCut is wrong or missing, telling me is a genuinely effective way to change it.

What the service does and doesn't commit to is written down in the Terms, and what happens to your data is in the Privacy summary. Both are written in plain language on purpose.

Who operates FirstCut