ClipForge: I built a machine that clips and posts short-form video while I sleep

Here's the thing nobody tells you about growing on short-form: it's a volume game, and the volume is inhuman. You don't win by posting one perfect clip. You win by posting constantly, across accounts, every single day, for a long time. Do the math on that by hand and you quit in a week. I know because I tried.

So I built the machine instead. It's called ClipForge.

What it actually is: an autonomous content engine with two tracks. One track takes long video — streams, long uploads, whatever — finds the clip-worthy moments, cuts them into formatted short-form clips, and posts them across accounts on a schedule. The other track runs a TikTok Shop play. Sitting on top is a Studio dashboard that runs the agency side: the briefs, the accounts, the output. The Engine underneath just... goes. That's the whole point. It's built to run unattended and post at volume, not to help me make one clip at a time.

Who it's for

Anyone whose growth depends on posting a lot of short-form and can't physically keep up — creators running multiple accounts, small agencies clipping for clients, brands trying to be everywhere at once. If your honest bottleneck is "I know what to post, I just can't post enough of it," that's the exact gap this fills.

Why I built it this way

Most "content tools" are really just editors with an AI button — they make the one clip slightly faster. That doesn't solve the actual problem, which isn't the single clip, it's the thousandth one. So ClipForge is architected as an engine, not an editor: it's a pipeline that keeps running, on Next.js and Postgres, designed so the boring, relentless part happens without me in the loop. The interesting work moved up a level — deciding what feeds the machine and what "good output" means — which is exactly where I want my attention.

It's built and running. What's next is tuning: better clip selection, tighter scheduling, and honestly just watching what it does at scale and fixing what breaks. No numbers to share yet — when there are real ones worth sharing, you'll get them here, same as always.

That's project number eight. The wall between "I wish I could post that much" and "the machine already did" is gone, and I'd rather live on the other side of it.