Scry: Programmable SQL search over 500TB internet index

What it does

Scry is a SQL query interface over a 500TB index of internet text. It indexes Reddit, Hacker News, LessWrong, arXiv, Stack Exchange, Wikipedia, and prediction markets—over 105 billion queryable rows. Users can write SQL or Datalog queries to search this data and integrate results into AI agents like ChatGPT or Claude. The service works as a plugin or API endpoint that agents can call to retrieve structured information from the indexed sources rather than relying on ranked link lists.

Who it is for

The tool targets AI agents and their users who need programmatic access to public internet data. Researchers, developers, and knowledge workers who ask questions the public record can answer are the primary audience. Integration is designed for non-technical setup: adding the plugin to ChatGPT or Claude takes about two minutes and requires no coding.

Pricing

Free for non-commercial use when capacity is available. Queries cost a fraction of a cent to a few cents when the service is under load. Pricing uses a congestion-based micro-auction model. Commercial use, custom data sources, or dataset work requires contacting the vendor directly.

How it stands out

Most search tools return ranked lists; Scry lets users run arbitrary SQL queries. This means questions like "every page mentioning compound X but not patent Y, grouped by domain since March" become single queries rather than manual filtering of search results. The indexed data covers forums, academic papers, market filings, and community discussions—sources that standard search engines may deprioritize. Integration with Claude and ChatGPT lets agents call Scry automatically as a tool rather than requiring manual interaction.

What a founder should check

First, verify the data freshness and completeness. The homepage states lag varies by source and is queryable, but a rival should measure whether staleness affects its competitive position, especially for time-sensitive data like market information. Second, examine switching costs and moat durability. Users build agents and workflows around Scry; how sticky is that integration, and what happens if a competitor indexes the same sources faster? Third, pressure-test the pricing model. Congestion-based auctions solve resource contention but may frustrate power users during peak hours, creating an opening for fixed-price competitors or unlimited tiers.

Thinking of building something like this?

Every launch here is a competitor to somebody's idea. If yours is close, check it against the market before you build: the Full Check names the rivals, the prices and the gaps.