What it does
Exa provides search infrastructure for AI agents and the developers building them. The site lists a Search API, a Contents API, an Agent API and a Monitors API, along with a library, integrations and an MCP server. A dashboard playground lets you run a query and view the results, an answer or structured output. Search can be tuned for speed or depth. The homepage shows a Fast mode at about 450 ms, an Auto mode at about 1 s and a Deep mode at about 10 s, with category options for the full web, news, companies, research, people and financials. A highlights feature returns short relevant excerpts instead of whole pages, which the site says reduces the tokens passed to a model. Deep search can return structured outputs that follow a schema you define, with grounded citations. The company says it trains its own retrieval models and runs its own index of web documents, and the site points to benchmark comparisons against other search providers. For enterprise use it lists SOC 2 Type II, GDPR and CCPA coverage, HIPAA safeguards with a BAA available, and zero data retention.
Capabilities
- Search API with Fast, Auto and Deep modes that trade latency for depth
- Category-focused search across web, news, companies, research, people and financials
- Highlights that return short, relevant excerpts to cut token use
- Structured outputs that follow a schema, with grounded citations
- Separate Contents, Agent and Monitors APIs
- Dashboard playground for trying queries
When to use it
- You are building an agent or assistant that needs current web results with citations
- You need structured company or people data pulled from the web through an API
- You want to reduce the tokens a model reads by passing short excerpts instead of full pages
- A coding agent needs documentation and repository context
Alternatives to explore
- Perplexity
- Brave
