PICopilot: An LLM-based Agentic Framework for Assisting Photonic Integrated Circuit Design via Script Generation

2026-08-03

AI Summary

This work addresses the inefficiency and high programming demands of photonic integrated circuit (PIC) design, which traditionally relies on manual scripting and struggles to meet growing complexity. To overcome these limitations, the authors propose the first large language model framework that integrates a multi-agent architecture with a customized retrieval-augmented generation (RAG) mechanism, enabling high-quality PIC design scripts to be automatically generated from natural language instructions. Evaluated on 48 diverse tasks, the approach achieves a 100% success rate—outperforming a generic RAG-enhanced GPT-5 by completing 21 additional tasks—while maintaining comparable latency and cost. This significant improvement in task completion and reliability enhances both the automation and maintainability of PIC design workflows.

Abstract

The rapid development of photonic integrated circuits (PICs) is shifting the design flow from traditional graphical user interface (GUI)-based methods to script-based methods for higher flexibility, portability, and maintainability. However, script-based design introduces new challenges, requiring designers to possess additional proficiency in tool application programming interfaces (APIs) and programming. It also demands greater effort and time because it is inherently less intuitive and more complex than GUI-based methods. As PICs grow in scale and complexity, the productivity gap between design needs and manual scripting capabilities continues to widen. To address this gap, we introduce PICopilot, the first large language model (LLM)-based agentic framework that assists in PIC design via automated design script generation from natural language instructions. PICopilot leverages a multi-agent architecture with a feedback mechanism and a specifically designed retrieval-augmented generation (RAG) pipeline, achieving a high success rate and reliability. Experimental results on a benchmark of diverse PIC scripting tasks demonstrate that PICopilot successfully completes all 48 tasks and outperforms other LLM-based approaches without incurring substantial extra latency or cost, even solving 21 more tasks than the advanced GPT-5 model with a general RAG pipeline.

Problem

• Photonic Integrated Circuit
• Script-based Design
• Productivity Gap
• Design Automation
• Large Language Model

Innovation

• LLM-based agentic framework
• script generation
• photonic integrated circuit design
• retrieval-augmented generation
• multi-agent architecture