Scholarlink

In recent years, large language models (LLMs) have demonstrated significant advancements in natural language understanding and generation. However, configuring these models to exhibit agent behaviors remains complex and resource-intensive, particularly for small and medium enterprises (SMEs) with limited technical capabilities. We introduce a novel configuration approach and framework designed to streamline the setup and deployment of agent-based LLMs within LangGraph. Our research objectives are to develop an accessible configuration scheme, provide an open-source parser to facilitate instantiation, and validate its practicality in real-world scenarios. The definition of agent behaviors, interactions, and environmental parameters is simplified by our framework, making it accessible even to non-experts. To evaluate the effectiveness of our approach, we conducted experiments with two distinct use cases: an automated customer service agent and an intelligent email bot. Our findings indicate that our framework significantly reduces development time, lowers technical entry barriers, and enhances the adaptability and performance of agent LLMs. The results demonstrate the potential of our solution in enabling SMEs to leverage advanced LLM capabilities with minimal technical overhead, bridging the gap between cutting-edge AI technologies and practical business applications.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Approach and Framework for Configuration of Agent-Based LLMs in Real-World Applications

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Jan-Philipp Schreiter,
  • Kirill Fuks,
  • Horst Hellbrück

摘要

In recent years, large language models (LLMs) have demonstrated significant advancements in natural language understanding and generation. However, configuring these models to exhibit agent behaviors remains complex and resource-intensive, particularly for small and medium enterprises (SMEs) with limited technical capabilities. We introduce a novel configuration approach and framework designed to streamline the setup and deployment of agent-based LLMs within LangGraph. Our research objectives are to develop an accessible configuration scheme, provide an open-source parser to facilitate instantiation, and validate its practicality in real-world scenarios. The definition of agent behaviors, interactions, and environmental parameters is simplified by our framework, making it accessible even to non-experts. To evaluate the effectiveness of our approach, we conducted experiments with two distinct use cases: an automated customer service agent and an intelligent email bot. Our findings indicate that our framework significantly reduces development time, lowers technical entry barriers, and enhances the adaptability and performance of agent LLMs. The results demonstrate the potential of our solution in enabling SMEs to leverage advanced LLM capabilities with minimal technical overhead, bridging the gap between cutting-edge AI technologies and practical business applications.