Updated
Updated · AiThority · Jul 23
System1 Launches Dogpile Fetch at $2 per 1,000 Calls as Study Finds 50% Search Gaps
Updated
Updated · AiThority · Jul 23

System1 Launches Dogpile Fetch at $2 per 1,000 Calls as Study Finds 50% Search Gaps

3 articles · Updated · AiThority · Jul 23

Summary

  • System1 rolled out Dogpile Fetch, a web search API and MCP server for AI agents that plugs into Claude Code, Cursor and other compatible clients in under two minutes.
  • Seven search APIs tested by Dogpile’s developer team showed any single index surfaced only about half of what the others collectively found, a gap the product aims to close by routing each query to the best backend.
  • Dogpile Fetch returns ranked results, knowledge graph data and People Also Ask prompts, while a Deep mode splits harder questions into multiple searches and deduplicates the results.
  • Pricing starts at $2 per 1,000 calls with free monthly credits and no credit card required, positioning the tool for developers building agent workflows.
  • Dogpile ties the launch to its metasearch roots: a 2006 study of 12,570 queries found 84.9% of first-page results appeared on only one engine, suggesting fragmented search remains a problem for AI agents.

Insights

Dogpile Fetch uses a protocol with known security flaws. Is better AI search worth the risk of a data breach?
Does piling up search results risk making AI agents more confused and less accurate, not more knowledgeable?