Animorph, a full-stack GEO platform company, said on Sept. 7 it unveiled querying.ai, an API for generative AI search optimization (GEO).

Querying.ai is a service that collects and provides the answers and cited sources shown on the AI search screen that users actually see.

Its feature is that it launches a browser directly and pulls the results shown on the screen, instead of using developer APIs provided by AI models. Because responses from a model API can differ from answers shown on the actual screen even for the same question, Animorph said it focused on the need for screen-based data to properly check brand exposure.

It also stressed that when a company sends a request including a question (prompt), a target AI service and a country, it can receive structured data not only for the answer text but also for the web search terms the AI actually ran to create the answer (search query fan-out), the sources cited in the answer and catalog information for products shown alongside it.

Up to 500 items can be bundled and sent in a single call, and results are received via webhook (automatic notification) or through a lookup method. Animorph handles all collection infrastructure, including browser automation and region and IP management, so companies can integrate immediately with a single API key.

Supported services include generative AI such as ChatGPT, Google Gemini and Perplexity, as well as Google AI Overview and AI Mode and Naver AI Briefing and AI Tab. It can respond in the same way to services that do not provide an API, and any service can be called using the same request and response format.

Song Jun-hyuk (송준혁), Animorph's CEO, said, "What companies need is not the answer itself that an AI model generates, but the search results that users see on screen." He said, "As how AI describes and recommends brands and products is becoming a new management metric, infrastructure is needed to repeatedly compare this data. Querying.ai's role is to help companies focus on interpretation and strategy instead of spending resources collecting data."