Quick recap
The whole argument in a few grounded points.
Speech + frames + evidence
AI Watch turns a video into a verifiable journey: every chapter, answer and note stays connected to the moment that produced it.
200 AI Processing Units a month—about 66 Standard Watch minutes • No card
Or read a real, unedited analysis of Apple's 76-minute WWDC 2026 keynote
Cited answer · 14:08
Which part changed between prototypes?
The outer grip geometry was widened while the central drive stayed the same.
Visual evidence
The exploded view isolates the changed housing.
AI Watch maps the video before it summarizes anything.
ChapterSpeech evidence identifies the original constraint.
Exact quoteThe frame reveals the part the speaker points to.
Frame + speechThe verified moment becomes a note, answer or self-test.
Notes + exportThe whole argument in a few grounded points.
Chapters with real frames and clickable timestamps.
Self-test questions and notes you can export.
Answers backed by stored evidence.
The gripper’s dimensions are on screen at 06:24 and are never said out loud. AI Watch reads the frame as well as the speech, so an answer can cite what it saw.




Get direct answers with exact timestamps and what’s on screen—no guessing.
A transcript gives you exact text. AI Watch adds visual understanding, chapters, real frames, study notes and evidence-led answers.
Yes. When captions do not exist, Watch interprets audio alongside the frames and labels spoken evidence accordingly.
Your AI Processing Units are refunded automatically. VidWords only keeps the charge when it delivers a finished analysis.
Yes—copy to clipboard or download as Markdown for Notion, Obsidian or a study document.