Why Proxycurl Shut Down, and What It Means for Your Pipeline
Proxycurl's product was a LinkedIn profile data API: send a profile URL or a name and company, and it returned structured JSON scraped from LinkedIn. That live-scrape capability was its differentiator, and it was also its single point of failure. In January 2025, LinkedIn sued Proxycurl's parent company, Nubela, alleging it used hundreds of thousands of fake LinkedIn accounts to scrape millions of profiles, including non-public data. Rather than fight a multi-year case against Microsoft as a bootstrapped company, Proxycurl settled and shut down the API on July 4, 2025. LinkedIn's complaint sought destruction of the scraped data; the settlement terms are confidential.
The lesson for anyone building a data pipeline is about architecture, not just one vendor. A provider whose entire product depends on a single scraping method can be taken offline by a single lawsuit or a single change to a website's defenses. When that happens, every integration built on it breaks at once, with no fallback. Proxycurl's own founder wrote afterward that LinkedIn data is still not safe to build on, and that legal does not mean safe.
A multi-provider waterfall is a fundamentally different architecture. Cleanlist enriches through 15+ licensed and aggregated data providers, so no single source, and no single sourcing method, is a point of failure. If one provider changes terms or goes away, the waterfall routes around it. That is the structural reason a licensed, aggregated model is the safer replacement for a single scraper. See the waterfall enrichment page for how the cascade works, and the API and MCP server for how to wire it into what you had built on Proxycurl.