Programmable internet search and indexing

AIThis description was written by a model from public signals. The facts behind it are listed under Evidence.

Problem

AI agents and builders depend on search APIs (Google Search, Tavily, Exa) that use opaque algorithms with fixed pricing, limited user control, and no way for communities to improve search strategies. These services struggle with resource contention and cannot handle the distribution of query hardness fairly.

Solution concept

A queryable internet index built on relational database infrastructure (ClickHouse) where users write SQL and Datalog queries directly against the index. Resource contention is managed through congestion-based micro-auction pricing rather than fixed fees, and the service is free when capacity is available. This gives users full control over search logic and transparent, efficient pricing.

Who has it

AI builders, agents, and information product creators who need to query large internet-scale datasets and want programmatic control over search logic.

How they handle it today

Today, they use proprietary search APIs like Google Search, Tavily, or Exa, which map their context to the provider's index with fixed algorithms they cannot see or customize. They pay fixed costs regardless of query complexity.

Why now

AI agents and large language models are becoming sophisticated enough to generate complex SQL and Datalog queries against large datasets. Frontier LLMs can now write the queries needed to extract value from a queryable index, and there is demand from AI builders for more control and transparency over search than proprietary APIs provide.

How it could charge

Usage-based pricing through congestion-based micro-auctions during high-demand periods; free for non-commercial use when capacity permits.