Custom Dataset on Demand for AI and Data Teams

Custom dataset on demand for AI and data teams: you name the public web data. We collect or buy it, clean it, and deliver it as CSV or API, with source address and date on every row.

A custom dataset on demand means you write the spec, we build the data. You skip the scraper, the blocks and the cleaning. Typical asks: a domain corpus for retrieval, a table of products or prices, a registry of companies or places, a stream of public posts on one topic.

What we deliver

input your spec: sources or kinds of source, fields, countries and languages, refresh
output one row per item, in your schema. Duplicates merged. Source address and fetch date on every row
format CSV, JSON file or API. You pick
rhythm once, or on a schedule
price on request

Sources and rights

  • Public pages and public registers: we load them without logging in. See data sourcing.
  • Bought datasets: we buy from vendors, then clean and merge them with other sources.
  • Some sources need a licence for commercial use. Reddit’s Data API Terms say commercial use needs a separate agreement with Reddit (Reddit). We tell you, per source, which route a dataset takes.

The Reddit case is not hypothetical: GummySearch closed because it could not get that licence (GummySearch alternative).

Sample

[TODO: one real sample dataset with 100 rows and its schema]

Start

Send the spec, even a rough one: a source, a field list, a country. We reply with what is possible, a sample and a price. Contact us. The general offer: custom data feeds.