# Etsy Enhances Image Captioning with Patronus AI's Multimodal LLM-as-a-Judge

> Discover how Etsy leveraged Patronus AI's cutting-edge technology to transform their image captioning capabilities.

**Source**: Patronus AI | **Published**: 2026-03-12 | **Type**: case_study

## Key Facts

- Etsy faced challenges with inaccurate autogenerated captions for product images.
- Patronus AI's Multimodal LLM-as-a-Judge provided a solution to mitigate caption hallucinations.
- The implementation led to improved caption accuracy and reduced errors.
- Etsy's AI team utilized the Patronus Experiments framework for continuous optimization.
- The collaboration resulted in enhanced user engagement and streamlined seller processes.

## Summary

\## Customer Background
Etsy is a global e-commerce platform specializing in handmade and vintage products, boasting hundreds of millions of items in its online marketplace. As a technology-driven marketplace, Etsy continually seeks innovative solutions to enhance the seller experience and improve customer engagement.

\## Challenge
Etsy's AI team aimed to leverage generative AI to autogenerate captions for product images uploaded by sellers. However, they encountered significant quality issues, as the autogenerated captions often contained errors and unexpected outputs, leading to a suboptimal user experience.

\## Solution
To address these challenges, Etsy implemented Patronus AI's Multimodal LLM-as-a-Judge, known as Judge-Image. This advanced AI solution evaluates image input to generate accurate text outputs, effectively reducing caption hallucinations and improving the overall quality of autogenerated captions. The Judge-Image utilizes a Google Gemini backbone, which has been shown to outperform other multimodal LLMs in terms of reliability and scoring distribution.

\## Implementation
The integration of Judge-Image into Etsy's existing systems allowed the AI team to iteratively measure and enhance the quality of their multimodal AI applications. By using the Patronus Experiments framework, the team could analyze performance metrics and refine their image captioning architecture over time. This iterative process involved evaluating images against captions and capturing confidence scores and reasoning for each decision made by the AI.

\## Results
The deployment of Patronus AI's solution led to a marked improvement in the accuracy of autogenerated captions. The Etsy AI team reported a significant reduction in caption errors and hallucinations, which in turn enhanced user engagement and streamlined the seller listing process. Although specific metrics were not disclosed, the qualitative feedback indicated a positive shift in both seller satisfaction and customer experience.

\## Quote
"Patronus AI's Judge-Image has transformed our approach to image captioning, allowing us to deliver more accurate and relevant content to our users. The insights gained from the Patronus Experiments framework have been invaluable in optimizing our AI systems." - Etsy AI Team

\## Conclusion
The collaboration between Etsy and Patronus AI exemplifies the power of innovative technology in enhancing e-commerce experiences. By leveraging the Multimodal LLM-as-a-Judge, Etsy successfully addressed its challenges with image captioning, leading to improved accuracy and user engagement. This case study highlights the importance of continuous optimization and the potential of AI to drive value in the e-commerce sector.

## Links

- [Read on Welcome.AI](https://welcome.ai/content/etsy-enhances-image-captioning-with-patronus-ais-multimodal-llm-as-a-judge)
- [Original source](https://www.patronus.ai/case-studies/etsy-leveraging-patronus-ais-multimodal-llm-as-a-judge-to-optimize-image-captionin)
- [Patronus AI](https://welcome.ai/company/patronus-ai): Featured company

---

Source: Welcome.AI | https://welcome.ai/content/etsy-enhances-image-captioning-with-patronus-ais-multimodal-llm-as-a-judge