📄 中文摘要
在构建零售库存 API 的第三部分中，作者分享了创建聊天机器人 API 的过程。经过前两部分的重构和迁移，零售 API 现已在生产环境中稳定运行，测试覆盖率达到 97%。接下来，作者开始构建聊天机器人 API，目标是实现一个能够记住对话内容、智能管理长对话并最终连接到零售库存数据的对话式 AI 服务。该项目使用相同的技术栈，增加了新的功能层。
从零开始构建聊天机器人 API：13 个 PR、许多故障以及一个真正有效的上下文窗口
📄 English Summary
Building a Chatbot API From Scratch: 13 PRs, a Lot of Broken Things, and a Context Window That Actually Works
In the third part of building a retail inventory API, the author shares the process of creating a chatbot API. After restructuring and migrating in the first two parts, the retail API is now stable in production with a 97% test coverage. The next step involves building a chatbot API aimed at delivering a conversational AI service that remembers what users say, intelligently manages long conversations, and eventually connects to retail inventory data. The project utilizes the same tech stack while adding a new layer of functionality.