Building BeforeNow in public

BeforeNow started from a small, annoying observation: the people with the most genuinely interesting footage are the worst served by editing tools.

Someone eight months into a rebuild, a stylist with two years of client work, a person renovating a house room by room. They have hours of real material on their phone. What they do not have is three hours to sit in a timeline and find the story inside it.

So the product is not another editor with more features. It is the opposite. You record, it cuts.

What that actually means

The bet is that the hard part of editing is not the interface, it is the judgement. Which clip is the turning point. Where the sentence ends so the cut does not slice through a word. Which four seconds of a nine minute session are worth keeping.

So BeforeNow reads the footage and listens to what you said, then builds an edit from that. Not a template with your clips dropped into slots, an actual edit decision list that responds to your material.

The reveal, the before and now moment with the gold divider, is the payoff the whole thing is built around.

What is shipped

In rough order:

• Capture. Recording progress footage over time, with an optional ghost overlay so your shots line up.
• Journeys. The footage organized as an ongoing story instead of a folder of clips.
• The auto edit. Scene detection and clip scoring in the render worker, so the good moments get chosen rather than the first ones.
• The reel engine. A speech-driven story edit: word-aligned cuts, kinetic captions, sound design, and three looks.
• The web editor. The same pipeline, running in a browser, free to try with clips you already have.

Every one of those took longer than expected, and the auto edit took roughly three times longer.

What is next

Motion design. Right now the edit gets the story right and the cuts land where they should, but the movement between shots is still generic. The next piece reads the transcript and directs the motion and sound effects to match what is being said, so a quiet line and a payoff line do not get the same treatment.

After that, the fuller editor for when the automatic version is close but not right.

Why in public

Two reasons, both selfish.

The first is that writing down what shipped forces me to be honest about what did not. It is easy to feel busy for a month and produce nothing that a user would notice.

The second is that this product only works if it makes something people are proud to post. That is not a thing I can judge alone from inside the codebase. If you try it and the reel is bad, that is the most useful thing you can tell me.

You can try it at beforenow.ai/create, and the app waitlist is on the home page.