Infineon Austria Deploys AI-Agent System to Cut Chip Design Workflow Hours

A master thesis by Hemalathaa Arumugam Elumalai, supervised by Konstantin Schekotihin at the University of Klagenfurt, presents the IFX Agent Ecosystem for Infineon Technologies Austria AG. The system uses two large‑language‑model‑powered agents: a Jira Agent that converts meeting minutes into structured Jira tickets, and a DocAgent that automatically generates technical documentation from source‑code repositories.

An ethnographic study of a 15‑person engineering team revealed that manual Jira ticket creation consumes about 390 hours per year and undocumented codebases add four to six engineer‑weeks per hire. In trials covering 376 production executions over 90 days, the Jira Agent achieved an 81.63 % composite score and delivered a 92.3 % time saving—approximately 360 engineering hours saved annually. The DocAgent scored 84.64 % across 17 repositories. Both agents integrate with Jira, Bitbucket and Confluence via an enterprise LLM gateway and are evaluated on a seven‑dimensional framework covering capability, reliability, robustness, safety, explainability, efficiency and user adoption.