E-commerce & ops
Price checks
Compare one product across stores, or pull a competitor’s whole catalog into a sheet.
› Nike running shoes under ₹5000 — store, title, price
No selectors. No scripts. Just a sentence.
Describe the data.
Get the spreadsheet.
Universal Extract reads public webpages and hands you clean rows and columns: prices, listings, directories, headlines. Download as CSV or JSON in seconds.
Try:
- No account needed
- Deleted after 30 minutes
- Checks robots.txt
- Never bypasses CAPTCHAs
books.toscrape.com → extract.csvExample output
20 rows · 5s · fetched without a browser| A | B | C | D | |
|---|---|---|---|---|
| 1 | title | price | star_rating | availability |
| 2 | A Light in the Attic | £51.77 | Three | In stock |
| 3 | Tipping the Velvet | £53.74 | One | In stock |
| 4 | Soumission | £50.10 | One | In stock |
| 5 | Sharp Objects | £47.82 | Four | In stock |
| 6 | Sapiens: A Brief History of Humankind | £54.23 | Five | In stock |
| 7 | The Requiem Red | £22.65 | One | In stock |
Prompt: “Book title, price, star rating, and availability” — one sentence, no selectors.
How it works
From a sentence to a spreadsheet, in three steps.
- 01
Describe it
Say which rows and columns you want, in your own words. Add a URL, or leave it empty and we’ll look up public pages.
“Laptops under ₹60,000 — name, price, rating, link”
- 02
We read the page
A plain fetch first, a real browser only when the page needs JavaScript. The model finds the repeating items and fills your columns.
✓ robots.txt allows
✓ fetched · 212 KB
● extracting 4 fields…
- 03
Download the table
Check the preview, then take CSV, JSON, or a Markdown table. Nothing is kept: results delete themselves after 30 minutes.
.csv.json.md
Writing a scraper
Universal Extract
Pick CSS selectors for every field
Write one sentence
Script breaks when the layout changes
Reads the page fresh on every run
Separate code per site
Same request works across sites
Hours to the first clean row
Seconds to the first clean row
Examples
Run one. Watch it work.
Each of these runs live against a real public page. Click, and the table fills in above.
- books.toscrape.comRuns live
Book catalog
Public demo catalog with title, price, rating, and availability.
Book titlepricestar ratingavailability - books.toscrape.comRuns live
Public product listing
A stable public catalog page. Title, price, and star rating.
Product titlepricestar rating - scrapethissite.comRuns live
Public directory
A simple public list: name, capital, population, and area.
Country namecapitalpopulationarea - news.ycombinator.comRuns live
News and research headlines
Public front page headlines with title and link.
Headline titlelink - web searchSearch mode
India price check
URL optional — search mode. Looks up public pages for a product and price.
titleprice Your own page
Any public listing, table, or catalog. Paste the URL and say which columns you want.
Start from scratch
Saved sourcesTeach it where to look
Save a page with a note about what it holds. When you leave the URL box empty, requests that match the note use that page.
Who it’s for
For people who need the table, not the scraper.
Analysts & founders
Research lists
Headlines, papers, launches, or funding rounds, as a sortable table instead of 40 tabs.
› Latest AI funding news — company, amount, date, link
Sales & partnerships
Directories
Public lists of companies, events, or places. No copy-paste, no cleanup.
› Every country — name, capital, population, area
Developers
Pipelines
One POST, one poll, one download. Stop maintaining a scraper per site.
› POST /api/jobs → completed → extract.csv
Responsible by default
Polite to the web. Forgetful with your data.
We built the limits in on purpose. They’re why teams can use this without a second thought.
- Public pages only
- We don’t log in for you or get around paywalls. If a page asks for a human check, we skip it and say so.
- robots.txt is checked
- Pages disallowed for all crawlers are skipped before we fetch them.
- Gone in 30 minutes
- Results live in memory and are deleted 30 minutes after they’re ready. Download what you need.
- Light on the sites we read
- One page at a time, a plain fetch first, and a browser only when the page can’t render without one.
API
Same result, three calls.
Sign up, create an API key in your dashboard, and run the same extraction from your own code.
curl -s -X POST http://localhost:4000/v1/jobs \
-H "Authorization: Bearer $UE_API_KEY" \
-H 'Content-Type: application/json' \
-d '{
"prompt": "Book title, price, and star rating",
"urls": ["https://books.toscrape.com/"]
}'Returns 202 with a job id. Leave urls empty to search public pages.
What table do you wish already existed?
Describe it. It’s free, and it takes a sentence.
Start extracting