📄 中文摘要
构建一个智能的个人AI助手是作者的目标，旨在超越传统的语音助手，能够真正理解用户的习惯、偏好和任务，并随着时间的推移变得更智能。然而，实际操作中遇到了许多挑战。作者深入研究了Alexa的云架构，试图理解其工作原理，以便为自己的项目奠定基础。通过这一过程，作者意识到，与其与大型语言模型（LLMs）抗争，不如使系统变得可预测，从而提升用户体验。
我尝试构建一个具有真实记忆的Alexa——三个月失败后的收获
📄 English Summary
I Tried to Build an Alexa with Real Memory — Here's What I Learned After 3 Months of Failure.
The author aimed to create a smarter personal AI assistant, one that goes beyond traditional voice assistants by genuinely understanding user habits, preferences, and tasks, and becoming more intelligent over time. However, the journey was fraught with challenges. The author delved into the cloud architecture of Alexa to grasp its workings and lay the groundwork for their own project. Through this process, it became clear that instead of fighting against large language models (LLMs), making the system predictable would enhance user experience.