1. Upload

PNG or JPG exports, one per slide. Slides are ordered by filename (so slide01, slide02, … works). If you add a transcript below and name each file with its time in HH-MM-SS form (e.g. 00-05-30.png for the slide at 5:30), the narration lines up exactly — extra text in the name is fine. If the names have no timestamps, that's OK too: with an API key the AI reads the transcript and places each slide where it's discussed; otherwise slides are spread evenly.

A VTT or SRT covering the whole session adds the narration under each slide (its end also sets the last slide's span). Leave it out for a slides-only guide — just the images plus any AI descriptions/alt-text you enable.

Optional. Used only for this run, never stored. A Parley key (sk-parley-…) is routed to MIT Parley automatically. Without a key the guide is built from the images + transcript alone.

2. AI features — optional, need an API key

Per-slide AI enrichments

3. Output

Working…