---
format: "aidr-story-markdown/v1"
id: "57cdde475b7ed50435b4bbc7b3af3347a83a41ecdb9824794e5acbcbdec78b82"
canonical_url: "https://aidr.today/57cdde47?lang=en"
title: "SteelAgent: LLM-Orchestrated, Physics-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-06T05:00:00.000Z"
category: "Research"
topics: ["agent"]
source_urls: ["https://arxiv.org/abs/2611.06953"]
summary: "SteelAgent couples metallurgical descriptors and machine-learning models with a language-model interface for property prediction. For gradient boosting, pooled out-of-fold $R^2$ fell from 0.859 on a random split to 0.113 when steel families were held out. Empirical coverage of nominal 90% prediction intervals ranged from 91.0% under random splitting to 37.6% on held-out sources."
---

# SteelAgent: LLM\-Orchestrated, Physics\-Informed Steel Property Prediction with Audited Generalization and Calibrated Uncertainty

> [Open the canonical story](<https://aidr.today/57cdde47?lang=en>)

**Published:** 2026-11-06T05:00:00.000Z
**Category:** Research
**Topics:** agent

## Summary

SteelAgent couples metallurgical descriptors and machine\-learning models with a language\-model interface for property prediction\. For gradient boosting, pooled out\-of\-fold $R^2$ fell from 0\.859 on a random split to 0\.113 when steel families were held out\. Empirical coverage of nominal 90% prediction intervals ranged from 91\.0% under random splitting to 37\.6% on held\-out sources\.

## Sources

- [Story source](<https://arxiv.org/abs/2611.06953>)