Context is all you need: Toward autonomous model‐based process design using agentic AI in flowsheet simulations

Abstract Agentic AI systems integrating large language models (LLMs) with reasoning and tool‐use capabilities are transforming software development, yet their application in chemical process flowsheet modeling remains largely unexplored. We present an agentic AI framework that provides assistance in an industrial flowsheet simulation environment. We show that GitHub Copilot (GitHub, Inc., 2026), using state‐of‐the‐art LLMs such as Claude Opus 4.6 (Anthropic, PBC, 2026), can generate valid syntax for our in‐house process modeling tool Chemasim from technical documentation and a few commented examples. Building on this, we develop a multi‐agent system that decomposes process development tasks: one agent solves the abstract engineering problem while another implements the solution as Chemasim code. We demonstrate the framework on typical flowsheet modeling examples, including a reaction/separation process, a pressure swing distillation, and a heteroazeotropic distillation with entrainer selection, and discuss limitations and future directions.