VAM
Find the right video. Cut the right moment.
A search-and-cut assistant for creators who work with lots of footage. Understands visuals, dialogue, on-screen text, and motion cadence. No cloud upload — scans and indexes locally on your disk.
HAI MỤC TIÊU
Two missions. Nothing else.
VAM is built around two questions — no feature is added unless it serves one of them.
Find the right video
From your prompt, VAM returns the top relevant results from your personal footage library. Not a web search — only what you've actually shot.
Cut the right moment
Within each video, VAM pinpoints the exact moment matching your prompt — down to the second — so you can drop it directly into your timeline.
TÍNH NĂNG
Every way you can describe a shot.
Each scene is analyzed across 8 layers. Ask anything; VAM answers in kind.
Visual semantic search
"Person jogging at sunrise" — VAM matches scenes with equivalent content, no manual tagging required.
Search by dialogue
Every spoken line transcribed. Tap on the "thank you" line to jump straight to that second.
Search by on-screen text
Logos, banners, burned-in captions — all indexed. Find any brand mark across hours of footage in seconds.
Motion-aware
Static, slow B-roll, hectic action sequence — VAM categorizes scenes by motion intensity.
Audio classification
Music, dialogue, silence, special FX. Filter for music-heavy scenes instantly.
Total privacy
Footage never leaves your machine. No upload, no server, no scene is used to train any model.