How Expand launched Expand Fetch on X

I'll verify Expand, the founders, and what Expand Fetch actually ships before writing the directory copy. The launch post and YC listing match. I'll pull the company page and announcement so the copy stays limited to verified facts. Expand Fetch is a fetch and crawl API for developers whose agents need answers from live web pages. Send a URL and Expand loads the page in a browser, then returns Markdown written by Iris, a vision model trained to read the rendered page and keep tables, icons, and layout. A second model, X-Ray, extracts useful JSON the page loads but does not show, such as prices, stock state, coordinates, or flight gates. Converters that drop those details are why an agent can miss a check mark on a pricing table and name the wrong plan. Every fetch is saved as a snapshot, and each line of Markdown links back to the place it came from, so a wrong answer can be checked against the page the agent saw even after the site changes. Highlights takes a question and returns only the passages that answer it. Expand says that cuts token use by more than 95 percent. Crawl runs the same extraction across a site, and WebSync reports pages added, changed, or removed. On MDBench, a private test Expand built with customers (1,094 questions on 326 hard pages), Claude Sonnet 5.5 answered 74.1 percent correctly from Expand's Markdown and 55.3 percent from the next API they tested, Exa. Fetch is live at $1 per 1,000 fetches. Expand is the San Francisco company behind the API, founded in 2024 by Tim Suchanek and part of Y Combinator's Summer 2024 batch. Suchanek was a founding engineer at Prisma and later founded Stellate, a GraphQL API management company Shopify acquired in 2024. He says the team hit this problem while building research agents, and that Fetch is the first step toward letting an agent query the web with a record of what it saw.