AI Pulse · Under the Radar

Keenable has released SELECT, a tool that lets people write SQL queries against the live web. Instead of pointing a person or a scraper at pages one by one, you can use custom functions such as WEB_SEARCH and SEM_EXTRACT, then ask an agent to return the results as structured rows.

That matters because SQL is already familiar to many analysts, operators, and data teams. A request like “find researchers who have moved between AI labs” becomes closer to a query than a bespoke research project. Early testers used SELECT for exactly that type of tracking, though Keenable has acknowledged accuracy caveats. The output still needs checking, especially where job titles, dates, affiliations, or similar names can be misread. You can see the early discussion in the Hacker News thread.

For a business owner, the interesting part is not SQL itself. It is the possibility of turning open-web research into a repeatable business process. Think competitor monitoring, supplier checks, market mapping, hiring intelligence, or finding changes across a set of company websites. The hard part will be deciding which questions are safe to automate and where a person must verify the answer before it drives a decision. This is the kind of thing we build into an AI command centre, where live research, internal data, and review steps can sit in one operating workflow.