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 market research report
Turn YouTube videos into market research
Paste review and demo videos. What buyers complain about, in their words — with a straight answer about what this cannot tell you.
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.
Market definitionDemand signalsPlayersPricing
Market research report
A structured read on a market: who is in it, what buyers say, how it is priced, where the gaps are.
Review videos, demos and walkthroughs are an unusually honest record of what breaks in real use, because a creator demonstrating a product on camera cannot skip the part that fails. Ancher pulls the recurring complaints, the pricing shown on screen, and the products named in comparison — while stating plainly that video volume is not demand and this report contains no market sizing.
The honest version
Video is excellent for pain points and useless for market size
The temptation with any scraped source is to count things and call it demand. View counts measure how good a thumbnail was. What review videos genuinely give you is language — the specific words buyers use for a problem, and the moment in a demo where a product visibly fails — and that is worth more for positioning than a number nobody can defend.
One second of one video, at 07:14, read two ways.Illustrative. The figure is invented; the gap it falls into is not.
...and this is where it falls over for most teams. Look at what happens when I
Failure shown on screen, never named aloud — not readThe specific never reaches the draft
Market research report
Failure shown on screen, never named aloud
Sync error, 3rd attempt
from 07:14
Lands in the draft, with its timecode
- Recurring complaints are counted across distinct videos and creators, not weighted by views.
- Pricing is read off the screen where it was displayed, with the date of the video attached, because pricing goes stale fast.
- The report states explicitly that it contains no market sizing, so nobody downstream mistakes it for one.
What comes out
Market research report, section by section
A structured read on a market: who is in it, what buyers say, how it is priced, where the gaps are.
Market definition
The market described in the words creators and their audiences use, rather than in a category name a vendor invented.
Demand signals
Complaints that recur across distinct videos, counted by how many separate creators raised them.
Players
Products named in comparison, with the criticism kept alongside the praise instead of only the verdict.
Pricing
Prices shown on screen, each carrying the video's date, and undated prices left out entirely.
Opportunities
Where the same complaint recurs and no named product addresses it, which is the closest thing to a gap this source can show.
YouTube market research 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 videos — in Ancher, or in whatever assistant you already use.
From these videos, build a market research report. (1) Define the market using the words creators and commenters actually use. (2) List recurring complaints with the number of distinct videos and creators raising each. (3) Name products only where discussed substantively, keeping criticism alongside praise. (4) Record pricing only where shown on screen, with the video's date; discard undated prices. (5) Identify complaints no named product addresses. State explicitly that this report contains no market sizing and that view counts are not demand.
The detail, if you want it
How each part of a YouTube video becomes a section
| From the YouTube video | Into | Why |
|---|---|---|
| Key frames from the video | Pricing | Pricing is nearly always displayed and rarely read aloud, so this section is empty without frames. |
| Caption track with timings | Demand signals | Complaint language is captured verbatim, because the exact phrasing buyers use is the most transferable output here. |
| Key frames from the video | Players | Comparison tables and side-by-side demos show competitors the narration never names. |
| Likes and comment counts | Demand signals | Comment volume on a complaint is a rough signal that it resonated, and it is reported next to the raw count rather than instead of it. |
| Title, channel, duration, date | Market definition | Publication dates bound the whole report, since a market read from year-old videos describes a year-old market. |
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 market research report. Ancher fills market definition, demand signals, players, pricing, opportunities 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
- Views are not demand. Any report that treats view counts as market size is misleading whoever commissioned it, and this one says so on its face.
- Sponsored reviews are common and not always disclosed in the audio. Check the description before treating a positive verdict as an independent signal.
- Creators cover what is popular to cover. A market read from video reflects what gets watched, which systematically under-represents unglamorous and enterprise segments.
Questions people ask
Is YouTube actually usable for market research?
For language and failure modes it is unusually good, because a demo cannot hide the step where something breaks. For anything quantitative it is a hypothesis generator only, and the report is written to keep that line visible rather than blur it.
How many videos do I need?
Enough that a complaint repeats across different creators. One reviewer's frustration is an anecdote; the same friction demonstrated by five independent creators is a finding worth acting on.
Can I combine this with other sources?
That is the stronger version. Add the vendors' own changelogs and pricing pages to the same collection, and any claim that contradicts what creators demonstrate on camera becomes visible immediately.
Bring your own YouTube videos
Everything you save lives in one workspace, so the market research report is built from your sources — not from a model's memory of the internet.
Open Ancher →Not a YouTube video? The same market research report also comes from PDF, X thread, LinkedIn post.