Turn one recording into a week of content

You recorded an hour. Somewhere in it are five moments worth posting, and finding them normally means someone scrubbing a timeline all afternoon and picking on instinct.

Endless does the reading. Make a transcript and the recording becomes text with speakers and word-level timestamps — searchable, quotable, and precise enough that a selection comes back with the exact seconds attached.

Then it can do the cutting. Ask for a moment as a vertical clip and Endless cuts that segment, reframes it to the shape the platform wants, burns the captions in from the transcript's own word timings, and files the finished .mp4 into Assets. Same conversation, no export step.

Make the transcript

Drag the file onto Transcripts, or paste a link — YouTube, Instagram, TikTok, X, or a direct link to a media file.

For a YouTube or TikTok link you get a choice, and it matters here. Fast reuses the video's existing captions and is ready quickly; High quality transcribes the audio instead, which is what adds speaker labels and exact word timings. Pick high quality — the word timings are what the cut snaps to and what the burnt-in captions are drawn from. Uploads and direct media links always run at high quality anyway.

Transcription is priced by the length of the recording, and the rate for each option is shown beside it; fast is several times cheaper. Nothing is charged for a transcription that fails, and pasting a link you have already transcribed opens the transcript you already have rather than paying for it twice.

Ask for the five

In chat, @ the transcript by name so Endless reads the right one instead of going looking.

Read @northsole-founder-interview and pick the five segments that would work best as Instagram Reels. For each: start and end timestamp, the hook line exactly as spoken, why this one works, and a caption. Aim the selection at reach — moments that stop a scroll in the first two seconds.

The reason attached to each pick is not decoration. It is what lets you argue with the selection, and arguing with it is how you aim the same hour somewhere else:

Same recording, different goal: five segments that would drive newsletter sign-ups rather than reach. I want the moments where the value is obvious and the explanation is missing, and tell me what to put on screen at the end of each.

Now optimise for engagement — the moments people quote or argue with. Rank them, and tell me which one I shouldn't post and why.

Three passes over the same hour give you three different weeks of content. The transcript is paid for once.

Then ask for the clip

The list of timestamps is the argument. The clip is the thing you post, and it is the next message rather than the next tool:

Cut the second one — the bit where he explains that focus means saying no. Vertical 9:16, keep him centred in frame, and burn in karaoke captions.

Now cut the other four the same way. Snap each one to the start and end of a sentence, and give me a square 1:1 version of the strongest for the feed.

Endless works in a sandbox for this: it loads the moment, cuts the segment, crops to the aspect you asked for — face-centred, so a talking head stays in frame rather than sliding off the side of a vertical crop — and burns the captions in. It runs as several steps, so it is not instant, and it reports back the way an editor would: the moment it picked and why, not the mechanics.

You get three shapes — 9:16 for Reels, TikTok and Shorts, 1:1 for the feed, 16:9 for everything else — and captions either as short cues or word-by-word karaoke, in a choice of fonts, colours and positions. Ask for a different moment, a different shape or a different caption style and it re-cuts.

What it does is cut, reframe and caption a moment. It is not an editing suite — there is no timeline of several clips, no transitions, no colour work and no audio mixing. One moment out of one recording, in the shape you need it.

Then adapt the same source

The clips are one output. The recording is a week of them, and it is all one conversation, so each of these is just the next message.

Now adapt it. For TikTok: rewrite those five hooks shorter and spoken-first, with three on-screen text options each that work without sound. For YouTube: the title and description for the full episode, chapters with timestamps, and three Shorts candidates that don't overlap with the Reels.

Turn it into an 800-word blog post in our voice. Keep the two strongest quotes verbatim and attributed to the speaker, and open on the moment at 14:20 rather than on the introductions.

Attach a project to the conversation first and "our voice" stops being a guess — the client's tone, their spelling, and the things they never say load with it.

Where the pieces end up

Each finished clip lands in Assets as an ordinary video, private to you until you share it, alongside the transcript and anything else the conversation wrote — a document, a web page, a PDF. A transcript is an asset like any other, which means it shares the same two ways: keep it to yourself, or share it with the workspace. See sharing a transcript.

If someone on your team would rather finish in their own editor, the transcript still exports SRT and VTT as well as plain text, speaker text and JSON. Those drop straight onto an editing timeline, so the edit gets the cue points and the captions rather than a list of numbers to retype — and renaming the speakers before you export carries the new names into every file.

What to check before you start

• Files must be under 2 GB. For a long show, split it or compress it first.
• One at a time. Recordings are transcribed in a queue and you can keep adding while it works; a failure leaves that item with a retry and the rest carry on.
• Transcripts can be switched off for a whole workspace, a team or one person — where it is off, the page is unreachable and chat cannot make one either. Creating one is a content action, so a Viewer cannot start it. See roles and permissions.
• Cutting can be switched off separately, because it is a skill rather than a feature. A workspace can transcribe but not clip, or the reverse.
• Chapters arrive a moment after the text does, and they are a fast way to find the shape of a long recording before you ask for anything.