The prompt
Score 10 of 10, Excellent
Use it when You recorded a webinar or talk and want the moments worth cutting into short vertical clips, with on-screen text and a post for each.
- {{platforms}}
- {{audience}}
- {{recording_info}}
- {{clip_length}}
- {{count}}
- {{full_link}}
- {{transcript}}
Full prompt and why it works
Prompt
Find the moments in the recording below that would work as short clips on {{platforms}}, for {{audience}}, so that our editor can cut them without watching the whole recording first. Each clip has to make sense to someone who never saw the full session. What the recording is (the event, the date, the speakers and their roles, who agreed to be clipped): {{recording_info}} Clip length we want: {{clip_length}} How many clips: {{count}} Link to the full recording: {{full_link}} The transcript, with timestamps: """ {{transcript}} """ Format: a numbered list of clips, each with: 1. Start and end time from the transcript, starting just before the first word and ending after the point lands. 2. Why it works alone: the idea, the strong line, the story or the disagreement it carries. 3. A hook for the first seconds: either the clip's own first line, or a short on-screen title if the clip needs setup. 4. On-screen text for the hook, under 6 words. 5. A caption for the post, under 60 words, naming who is speaking and linking to the full recording. Then rank the clips by how well they stand alone, and list moments that are good but need context, so we can use them as quote posts instead. How to choose: - Prefer moments with a clear claim, a specific example, a number the speaker gives, or a short story with an ending. - Skip greetings and housekeeping, slides being read out, inside jokes, and stretches that rely on "as I said earlier". - Never cut a sentence so that it changes what the speaker meant; read the lines just before and after. A good clip starts on the point, not the run-up to it, and the speaker would be happy to see it shared. For example, a start note in the right form: "Start at 14:02 on 'The mistake we made', not at 13:50 where the speaker sets it up." That example shows the form only. Quote the transcript exactly in on-screen text and captions, and do not add words the speaker did not say. If the transcript has no timestamps, give the opening and closing words of each clip and mark it [ADD TIMES]. If a speaker's consent to be clipped is not stated, add [CHECK CONSENT] to their clips.
Variables
- {{platforms}}
- Where the clips will go, for example "LinkedIn and Instagram Reels".
- {{audience}}
- Who the clips are for, for example "HR managers at mid-size companies".
- {{recording_info}}
- The event, date, speakers and their roles, and who has agreed to be clipped.
- {{clip_length}}
- How long each clip should be, for example "30 to 60 seconds".
- {{count}}
- How many clips you want, for example 6.
- {{full_link}}
- Where people can watch the whole recording, or "not published yet".
- {{transcript}}
- The full transcript with timestamps, from your recording or captioning tool.
Why it works
- Who it's for. The output is a cut list for an editor: exact times, why each clip stands alone, its hook and its caption, so nobody has to watch the whole recording.
- What to avoid. Words are quoted exactly and no sentence is cut so that it changes the speaker's meaning, which protects the speakers and your credibility.
- If unsure. Missing timestamps are marked [ADD TIMES], and any speaker whose consent is not stated gets [CHECK CONSENT] on their clips.