Overviews (AI)

A timestamped, collapsible breakdown of a single video — jump straight to the part you need.

An Overview is a timestamped, collapsible map of a single video: the sections it moves through, the topics inside each one, and the exact moment each begins. It's how you find the four minutes you actually need in a 90-minute lecture.

Using it

  1. 1
    Click Overview on any video row, in the popup's course map or in the in-page panel.
  2. 2
    Wait a few seconds. The tree appears in a panel under that row, with a count line (9 sections · 31 topics), a search field, and a Copy button.
  3. 3
    Click any timestamp. On the video you're currently watching this seeks the live player. On any other video it opens YouTube at that moment.

It's cached

Once built, an Overview is saved — the button turns green (“Overview ready”) and opens instantly forever after, on every surface, without spending anything. Popular videos are often already built by someone else, in which case yours is free and instant too.

You don't need a course first

On a watch page for a channel you haven't coursified, below the “Coursify this channel” button sits a This video card with the same Overview button — an Overview is about one video, not a channel. It's saved against that video, so if you coursify the channel later, its row already reads “Overview ready”. The limits are unchanged: on our shared keys it still counts toward that channel's free Overviews and your daily limit. One video, no course →

Sharing timestamps

Above the toolbar in the in-page panel is a Share timestamps in comments row. It produces the same timestamped outline as Copy, formatted for a YouTube comment — chapter markers that other viewers can click. Creators find this genuinely useful, and what's copied and what's posted can never drift apart because both come from the same source.

Copy the whole outline

One click puts the entire timestamped tree on your clipboard as plain text — drop it in your notes, a doc, or a comment.

Search inside it

The search field filters sections and topics as you type, so a long video's outline collapses to just the bit you're after.

How it's built

Two different methods, tried in order. You never have to choose.

  • 1 · From the video's own captions (almost always). CoursifyYT reads the caption track in your browser and hands it, with its clock, to your AI provider. Because the model is reading real timestamps rather than guessing them, the anchors line up with the video exactly.
  • 2 · Gemini watches the video (fallback). For videos with no caption track — or too little speech to be worth summarizing — Gemini processes the video itself. Slower and rarer, but it means music-heavy and caption-less videos still work.

You can tell which one ran

A caption-built Overview shows a “Built from this video's captions · Read the transcript” footer, which flips the card straight to the full Transcript. Gemini-watched trees don't show it — there's no transcript to offer.

Limits

Free (our keys)Your own key
Per course3 distinct videosUnlimited
Per day6 generations per IP addressUnlimited
Already-cached overviewsFree and unlimitedFree and unlimited

Either key lifts the cap — your main AI key or the optional Gemini fallback key, since between them they cover both methods. Adding a key →

Two different limit messages

“You've used your 3 free overviews for this course” is the per-course cap — adding a key is the only thing that changes it, so there's no retry button. “Daily free Overview limit reached” is the per-day one; it resets at midnight UTC.

A note on accuracy

Overviews are AI-generated from what the video says. They're reliable for navigation — finding where a topic starts — and should not be treated as a substitute for watching when precision matters. If an outline stops well before the end of a long video, generate it again; a truncated tree is detected and never cached, so a second attempt rebuilds it rather than re-serving the stub.