Exa (formerly Metaphor) is a search API that uses neural embeddings to find documents semantically similar to a query, rather than matching keywords. Ask for 'companies doing interesting work in materials science startups' and it finds documents that actually discuss that, not just pages containing those words. The API is built for developers and AI teams who want to give their AI applications access to real-time web knowledge. Use it to ground LLM responses in current sources, build research agents, or power semantic discovery features. The contents endpoint returns full page text rather than just URLs. For AI developers building retrieval-augmented applications that need to search the web by meaning rather than keyword, Exa's neural search capability enables use cases that traditional search APIs can't handle.