Tutorial Videos as Code with HyperFrames and Claude

Rebuilding DoveRunner's product tutorial videos as code — HyperFrames compositions, real Console captures, synthesized narration and captions, written and maintained with Claude — so a UI or docs change means a re-render, not a re-shoot.

DoveRunner’s integration tutorials on YouTube predated the rebrand from PallyCon, and the Console they showed had moved on. Re-recording, re-editing and re-voicing every video each time the product changes doesn’t scale, so I treated it as a maintenance problem, not a production one.

Each tutorial is now a project in one repository: a tutorial spec and narration script checked against the current docs, real Console captures with redactions and measured highlight boxes, and scenes built as HyperFrames HTML compositions on a shared component library. Narration is synthesized and caption-aligned locally, a music bed is carved to the timeline, and the whole video renders deterministically. Claude does most of the scene building, timing and tooling; a human approves the narration before any voice is generated and the technical accuracy before anything is published.

What started as a single-video proof of concept in September 2026 is now the production path. Tutorials on CLI packaging, license tokens, playback testing, DRM packaging with AWS Elemental MediaConvert and MediaLive/MediaPackage, Transcoding & Packaging and forensic watermarking are published as lessons in DoveRunner Lab.