Voodoo.tube watches a long-form video, finds the moments worth clipping, and ships captioned shorts scored for virality. We built the platform end to end, and it was later acquired.
Creators and podcasters sit on hours of long-form video while their audience lives on shorts. Bridging the two is brutal, manual work: watch the whole thing, find the moments that pop, cut them to vertical, caption them, and hope you guessed right.
Voodoo.tube set out to make that judgement call on its own, reading a video well enough to surface the clips worth posting rather than chopping it into arbitrary pieces.
Drop in a video URL and the platform handles the rest. It transcribes the audio, analyzes the content, and pulls the high-potential moments into finished vertical clips, each one captioned and rated.
Demo: upload to analysis to finished, scored clips
At its peak, the platform was turning thousands of long-form videos into short-form every month, each one clipped, captioned and scored without a human in the loop. Then it sold.
Voodoo.tube grew into a platform processing thousands of videos every month, with generation time running at roughly half a clip's source length, so a one-hour video turned around in about thirty minutes.
It worked well enough to be acquired, and the product is still live at voodoo.tube today. This is the outcome we build for: software that earns its place, then outlives the engagement.
A conversational travel-disruption claim assistant that helped more travelers finish the flow.
A four-pass Claude pipeline that turned a week of manual scripting into same-day turnaround.
Want a build like this? Start with a software audit, see how we build, or read the guide to building software without a tech team.
Next / Your build
Taking on new builds
Tell us what you're making. We reply within a day with a fixed price and a date.
