Keenable SELECT
An AI agent tool that exposes web search interfaces through a SQL-like query syntax, allowing developers to query internet information using standard SQL statements. An underlying AI agent handles search execution, page scraping, and structured data extraction. The product is positioned as a programmable search and data pipeline tool.
Timeline (last 90 days)
Keenable SELECT is a product that turns "searching the web" into SQL statements, treating the internet as a queryable virtual data table
Keenable SELECT received 48 points and 12 comments on Hacker News
Using SQL as a search interface brings three benefits: structured results, strong composability, and developer friendliness
In Keenable SELECT, SQL is only the interface layer, and the actual search is executed by an AI Agent behind the scenes
Mapping messy web content into well-organized data tables is the technical core and challenge of such products, and is where large models come into play
The architecture of Keenable SELECT is conceptually similar to virtualized Table-Valued Functions and PostgreSQL's Foreign Data Wrapper (FDW) mechanism
Keenable SELECT directly reaches the public internet with each query, bypassing index construction; its timeliness is superior to traditional RAG, but it loses the semantic recall and cost-amortization advantages of vector indexes
Reliably mapping web pages into a fixed schema is a longstanding difficult problem in the NLP field, and errors in field extraction, omissions, and hallucinations can contaminate query results
In structured query scenarios, hallucinations can appear in the form of numbers that seem plausible but are actually incorrect, and once they enter a data pipeline they propagate silently with extremely high investigation costs
Such products should establish source traceability mechanisms, with each returned field accompanied by precise original web page location (URL + text offset) for independent downstream verification
1 more timeline events
All Facts (10)
Keenable SELECT is a product that turns "searching the web" into SQL statements, treating the internet as a queryable virtual data table
50%UnverifiedUsing SQL as a search interface brings three benefits: structured results, strong composability, and developer friendliness
50%UnverifiedIn Keenable SELECT, SQL is only the interface layer, and the actual search is executed by an AI Agent behind the scenes
50%UnverifiedMapping messy web content into well-organized data tables is the technical core and challenge of such products, and is where large models come into play
50%UnverifiedThe architecture of Keenable SELECT is conceptually similar to virtualized Table-Valued Functions and PostgreSQL's Foreign Data Wrapper (FDW) mechanism
50%UnverifiedKeenable SELECT directly reaches the public internet with each query, bypassing index construction; its timeliness is superior to traditional RAG, but it loses the semantic recall and cost-amortization advantages of vector indexes
50%UnverifiedReliably mapping web pages into a fixed schema is a longstanding difficult problem in the NLP field, and errors in field extraction, omissions, and hallucinations can contaminate query results
50%UnverifiedIn structured query scenarios, hallucinations can appear in the form of numbers that seem plausible but are actually incorrect, and once they enter a data pipeline they propagate silently with extremely high investigation costs
50%UnverifiedSuch products should establish source traceability mechanisms, with each returned field accompanied by precise original web page location (URL + text offset) for independent downstream verification
50%UnverifiedTurning search into SQL allows searches to be called programmatically, triggered by cron scheduled tasks, and embedded into automated workflows (such as Airflow)
50%