If you record videos regularly and end up with a messy clips folder, I built a macOS app to help. AlgoMommy organizes raw video files into an existing folder hierarchy based on short spoken instructions you say inside the recording (e.g. “Hey Cleo, this is wedding B-roll - tag it green” or “Hey Cleo, this is a New York sunrise”). After recording, drop the clips into AlgoMommy; it listens for the instructions you said during recording and copies the files to the appropriate sub-folder. Download (DMG): - https://download.algomommy.com/app/macos/public/AlgoMommy-1.... - 492 MB - most of this is WhisperKit, signed & notarized - sha256: de354d073467aece719433f170f1886c5ab781835e4caccc4054827ee3320ee5 Demo video (48s): - Youtube: https://www.youtube.com/watch?v=Ucj-D9yOKBc - Raw MP4: https://download.algomommy.com/public/AlgoMommy.mp4 Try it now: 1) Install the DMG and log in with this shared demo account: 3) Drag & drop a few videos and click "Organize videos" Auth is required because the service uses paid LLM calls. What leaves your machine / privacy: - No raw video or audio is uploaded. - Audio is extracted + transcribed locally. - The only data sent is: (1) ~30s text transcription snippets for segments that appear to address “Cleo”, and (2) the list of relative destination folder paths under the root you selected (so the model can choose a folder). - Those snippets are used to decide destination folder + tags/metadata. Why “Cleo” (and why natural language): - WhisperKit is accurate but slow, Apple's SpeechAnalyzer is fast but less accurate. AlgoMommy uses SpeechAnalyzer to quickly find "Cleo"-addressed segments, then re-transcribes just those ~30s chunks with WhisperKit for accuracy. Safety: - AlgoMommy only "copies" videos, it never "moves" / "overwrites" videos. Inspiration: - Initially this was a Python script for my partner, to solve this very issue. I subsequently decided to "productize" the solution as a macOS app. I’m looking for feedback on: 1) Does this solve a significant problem for you? 2) Is the workflow clear and robust in real-world usage? 3) Today you must pre-create the folder hierarchy: is that acceptable, or should the app be able to create target folders in some cases? If folder creation is valuable, what rules would make it safe/predictable? 4) Any voice commands you’d want beyond “organize/tag this clip” (e.g. “start a new video file here”)? Happy to answer questions and go deep on implementation details if anyone’s interested. |