AI Agents for Photonic Integrated Circuit Design Automation

TL;DR

This paper introduces PhIDO, a multi-agent framework utilizing large language models to automate photonic integrated circuit design from natural language, achieving up to 91% success in simple cases and highlighting future automation steps.

Contribution

The paper presents a novel multi-agent framework that leverages reasoning large language models for converting natural language PIC design requests into layout files, with comprehensive evaluation.

Findings

Up to 91% success rate for single-device designs.

Gemini-2.5-pro, o1, and Claude Opus 4 achieved ~57% success on small component counts.

Gemini-2.5-pro required fewer tokens and was more cost-effective.

Abstract

We present Photonics Intelligent Design and Optimization (PhIDO), a multi-agent framework that converts natural-language photonic integrated circuit (PIC) design requests into layout mask files. We compare 7 reasoning large language models for PhIDO using a testbench of 102 design descriptions that ranged from single devices to 112-component PICs. The success rate for single-device designs was up to 91%. For design queries with less than or equal to 15 components, o1, Gemini-2.5-pro, and Claude Opus 4 achieved the highest end-to-end pass@5 success rates of approximately 57%, with Gemini-2.5-pro requiring the fewest output tokens and lowest cost. The next steps toward autonomous PIC development include standardized knowledge representations, expanded datasets, extended verification, and robotic automation.