This research demonstrates how to deploy effective language agents in resource-constrained environments for specialized domains and less-common languages.

• Context-efficient architecture reduces token consumption through hierarchical section search
• Scenario-based dialogue generation creates comprehensive training data with minimal human effort
• Domain specialization in Korean chemical toxicity information despite limited resources
• Practical implementation balances performance with operational constraints

This work matters for medical contexts where specialized information must be accessible in local languages despite computing limitations, enabling improved chemical safety information dissemination in healthcare settings.

Building Resource-Constrained Language Agents: A Korean Case Study on Chemical Toxicity Information