Andrej Karpathy — Intro to LLMsY Combinator — How to get startup ideas
Your YouTube video
Paste a YouTube link. Nothing to upload.
←YouTube video
Make a social post pack
Turn a YouTube video into social posts
Paste a link. Hooks built from the sharpest moment in the video, in your voice, with credit attached.
No link handy? Try one:
Free to try, no card. Your YouTube video opens in Ancher.
Caption track with timingsKey frames from the videoDescription and linked sources
Kept in parts
Every one of these stays separately addressable instead of collapsing into one block of text.
Hook variantsThread bodyQuote cardsCall to action
Social post pack
Publish-ready posts in your voice, with hooks, a body, quote cards and credit.
The shareable part of a video is almost never its thesis — it is one specific number, one named example, or one moment where the speaker reverses an expectation. Ancher finds those moments, builds hooks around them rather than around the topic, and pulls the frame they appeared on so a post can carry the visual that made the point.
The honest version
The thing worth posting is usually a frame, not a line
Posts that travel carry an image people can read in one second — a chart, a comparison, a number on a slide. A tool working from the caption track can only give you sentences, so its output reads like a summary in five parts and performs like one. The frame is the asset, and it has to survive ingest to be usable.
One second of one video, at 11:47, read two ways.Illustrative. The figure is invented; the gap it falls into is not.
...everyone assumes it goes up here. It doesn't. It goes to and stays there.
The chart everyone screenshots — not readThe specific never reaches the draft
Social post pack
The chart everyone screenshots
-38% by week 6
from 11:47
Lands in the draft, with its timecode
- Frames are captured, so a post can carry the chart the creator put on screen instead of describing it.
- Hooks are generated from specific moments — a number, a name, a reversal — rather than from the video's overall subject.
- Credit to the creator is generated with the post, which is both decent and the reason creators amplify rather than complain.
What comes out
Social post pack, section by section
Publish-ready posts in your voice, with hooks, a body, quote cards and credit.
Hook variants
Three variants built on the single sharpest moment, not on the video's theme, because a theme does not stop a scroll.
Thread body
A sequence that opens on the payoff even if it arrived at minute forty, then earns it backwards.
Quote cards
The specific frames worth attaching, named by timecode so you can pull them yourself if you prefer.
Call to action
Usually an open question the speaker left unanswered, because a question draws replies where a link draws nothing.
Source credit
Creator handle and video link, generated with the post rather than left as something to remember.
Video-to-social-posts prompt
The link on its own gets a summary. This is the instruction that produces the 5 sections above, in that order, with the rules that keep them honest. Paste it with your YouTube video — in Ancher, or in whatever assistant you already use.
From this video, write a social post pack. (1) Identify the five most specific moments — those containing a figure, a named example, or a reversal of expectation — with timecodes. (2) Write three hook variants for the single strongest one. (3) Write a thread that opens on that moment rather than on the video's thesis. (4) Name the frames worth attaching as images, by timecode. (5) Write one credit line naming the creator and linking the video. Match the voice of my saved writing; do not restate the video's summary.
The detail, if you want it
How each part of a YouTube video becomes a section
| From the YouTube video | Into | Why |
|---|---|---|
| Caption track with timings | Hook variants | The sharpest moments are found by scanning for specificity — a figure, a proper noun, a contradiction — not by summarising. |
| Key frames from the video | Quote cards | A post carrying the original chart consistently outperforms one describing it, and the chart only exists in the frames. |
| Caption track with timings | Thread body | Caption timings give the sequence of the argument, which the post then inverts so the payoff comes first. |
| Likes and comment counts | Hook variants | Comment volume marks which claim the audience argued about, which is the most reliable available signal for a hook. |
| Description and linked sources | Source credit | Channel and video URL come from the source metadata, so attribution is automatic rather than optional. |
Everything Ancher keeps from a YouTube video
Caption track with timings
Published and auto-generated captions both come through, and every line keeps the second it was spoken.
Key frames from the video
Frames are sampled and scored, so a number shown on a slide survives even though it was never said out loud.
Description and linked sources
The creator's own cited links come along, so a claim can be traced past the video.
Likes and comment counts
Reception is a weak signal about correctness and a strong one about which claims got attention.
Title, channel, duration, date
Basic provenance, kept so the finished work can attribute the video properly.
The four steps
- Add the YouTube video. Paste a YouTube URL. Ancher pulls the caption track with its timings, samples key frames from the video itself, and keeps the description links and engagement counts on the same source.
- Nothing gets flattened. Caption track with timings, key frames from the video, description and linked sources, likes and comment counts, title, channel, duration, date stay separately addressable instead of collapsing into one block of text.
- Ask for a social post pack. Ancher fills hook variants, thread body, quote cards, call to action, source credit using the mapping above, not a generic summary pass.
- Check it before you use it. Every specific keeps a link back to where it came from in the YouTube video, so a claim can be verified without reopening the source.
Where this goes wrong
- Posting a creator's thesis as your own is the fastest way to get called out. Lead with the specific and credit in the same post, not three posts later.
- A thread that follows the video's running order buries the payoff. The strongest moment goes first even if it came at the end.
- Frames from a paywalled or unlisted video should not be posted. Use public frames from a public video, and check the creator's stated reuse terms if you are unsure.
Questions people ask
Will it sound like me or like AI?
Ancher drafts against the writing already in your workspace, so voice comes from your own corpus rather than from a tone setting. That usually makes the first draft closer than a generic tool's fifth, though it needs your saved writing to work from.
Is repurposing someone else's video acceptable?
With credit and your own commentary, yes — that is most of how social works. Reposting their frames with no attribution is not, and the pack generates the credit line so the decent version is also the easy one.
Can I get LinkedIn and X versions together?
Yes. The same extracted moments feed both; hook density and length differ between the two, but the underlying claims and the credit line stay identical so you are not maintaining two versions of a fact.
Bring your own YouTube videos
Everything you save lives in one workspace, so the social post pack is built from your sources — not from a model's memory of the internet.
Open Ancher →Not a YouTube video? The same social post pack also comes from PDF, X thread, LinkedIn post.