Speaker
Description
The growing complexity of telescope configurations, such as those in the Cherenkov Telescope Array Observatory (CTAO), demands efficient and reliable tools for data modeling and validation. This project presents a multi-agent application, called 'CTAgent', designed to automate the generation of Pydantic Python models directly from free-text descriptions or structured files. It also explores the use of the multi-agent framework AutoGen, as well as minimal function tools available in new OpenAI models. The system incorporates a feedback loop to verify and refine generated code before user presentation, streamlining the workflow for astrophysical data management. As an additional investigation, benchmark tests were conducted on two OpenAI large language models (GPT-o3 and GPT-5) to evaluate their reasoning capabilities in the niche of astronomy using the TevCat Catalogue. They were each tested on low and high reasoning efforts.
