expand.ai

expand.ai is a developer tools and infrastructure company that converts websites into type-safe APIs returning structured data. Users pass one or more source URLs to the service, which can auto-generate a data schema for the target pages or accept a user-defined schema, and then returns typed records matching it. Alongside instant on-demand requests, the company offers dataset building, in which extraction results are synced into a dataset that can be exported to destinations such as S3, Postgres or Google Sheets; this dataset capability is labeled as coming soon on the company website.

Beyond scraping public web pages, the product accepts user-supplied inputs such as internal documents and PDFs, and can use the wider internet as a data source in response to a natural-language prompt. A separate markdown endpoint returns semantic markdown intended for direct use as context for large language models. The company states it handles operational aspects of web extraction on the customer's behalf, including bot protection, JavaScript rendering, stealth mode, proxies, browser infrastructure and auto-healing crawlers, and that extracted data is checked and traced back to its source.

Founding story

Founder and CEO Tim Suchanek had worked on developer tools for about eight years before starting expand.ai, serving as a founding engineer at Prisma and then founding Stellate. While building AI applications he found that obtaining the right data was frequently the blocker to starting a project, which led him to build expand.ai so that the web could be used as a data source for AI applications.

Business model

expand.ai sells extraction and dataset infrastructure as an API-based developer service; access at the time of the sources was gated behind a waitlist.

Traction

The company website reports 22 million pages scraped to date. Y Combinator lists the company as active with a team size of two.

Latest developments

The website advertises a forthcoming dataset feature that persists extraction results into named datasets queryable through a client and exportable to external stores, and access to the product is offered through a waitlist.