如何使用 Amazon Bedrock 和 Lambda 构建无服务器 AI 代理
📄 中文摘要
为了满足内部 HR 问题的需求，开发了一种工具，能够回答关于休假余额、政策查询和团队日程等问题。传统的聊天机器人依赖于大型语言模型（LLM），但其容易产生虚假答案，因此需要访问真实数据。Amazon Bedrock Agents 通过允许 LLM 通过 AWS 的“操作组”调用后端函数来解决这一问题。LLM 读取函数描述，判断哪个函数与用户问题匹配，提取自然语言中的参数并调用相应函数。整个过程无服务器运行，无需管理 EC2、容器或服务器。通过本教程，用户将构建一个由 Lambda 函数支持的 Bedrock Agent，处理四个 HR 操作：检查休假余额、提交请假请求等。
📄 English Summary
How to Build a Serverless AI Agent with Amazon Bedrock and Lambda
An internal tool was developed to address HR inquiries, such as leave balances, policy lookups, and team schedules. Traditional chatbots rely on large language models (LLMs), which can generate inaccurate answers due to a lack of access to real data. Amazon Bedrock Agents resolve this by enabling LLMs to call backend functions via what AWS refers to as 'action groups.' The LLM reads function descriptions, determines which function aligns with the user's question, extracts the necessary parameters from natural language, and invokes the function. This entire process operates serverlessly, eliminating the need for EC2, containers, or server management. By the end of this tutorial, users will have a functioning Bedrock Agent supported by a Lambda function that manages four HR operations: checking leave balances, submitting time-off requests, and more.