In 2024 our founders were running outbound for a B2B company and spending more time maintaining scrapers than talking to prospects.
Each new source meant a new script, and every site redesign broke it. They started building an agent that could work out the structure of a page on its own.
The internal tool became the product. Today Cyclosa runs contact extraction for sales teams, recruiters and lead-gen agencies who need data from thousands of sites they have never seen before.
Make public web data available to any team without writing a line of scraping code.
Finding contacts should take minutes, not an engineering sprint. Every scraping tool before us asked for a parser per site, and every redesign broke it. We put the work of reading a page onto the agent instead, so our customers can spend their time on the conversations that follow rather than on maintaining extraction scripts.
One agent that can extract any structured data from any site, on request.
Contacts are the first data type. Prices come next, then whatever a customer defines. We are building toward a single engine that adapts to the spec instead of the source, so adding a new kind of output becomes a configuration question rather than an engineering project.
The rules the agent runs by, on every job, for every customer.
We extract information that is already published and openly accessible. No credential walls, no private accounts, no data a visitor could not see themselves.
Each job reads the domain you submitted and a handful of its contact pages, nothing beyond it. Fast results should not come at the cost of someone else's infrastructure.
Results belong to the account that ran the job. We do not resell extracted records or build a shared contact database from customer runs.
Eleven people in San Francisco building one crawler for every website.
Questions about coverage or volume pricing: we answer within one business day.
Subscribe, add the sites you care about, and get contacts back in minutes.