Over the Summer, I worked as an intern for Costco Logistics, where I spent the majority of my time developing an AI-driven project for their website and services. This experience gave me a firsthand insight into how major corporations are integrating artificial intelligence into their daily operations, and it shaped my understanding of the growing role that AI plays in modern logistics. In my presentation, I'll be sharing what I learned throughout this process, from technical skills that I gained to the broader trends that I observed. Using the project that I worked on as a case study, I'll walk through the full lifecycle of building an AI tool: designing the model, setting up the database, editing instructions, training the LLM, and making the bot user-friendly. Along the way, I’ll explain the challenges I encountered, the methods I used to optimize the chatbot’s performance, and the work I completed on both the front-end and back-end of the AI bot to create a functional, reliable tool. In addition to detailing the process of the project and its results, I will discuss how it and other novel AI endeavors transform the shipping industry around the world. Ultimately, my goal is to show how real-world AI development works inside a major corporation like Costco and why these technologies are becoming essential to industries across the country.
The purpose of this project was to create an AI chatbot that stored Costco's depot information and instructions regarding order delivery and shipping. The initial aim was to provide truckers and shipping companies with a tool to use for Q&As instead of contacting tech support; however, its functions eventually expanded beyond this task. I started by creating the chatbot over a week and using a pre-set knowledge base from Google Gemini. Throughout the next few weeks, I entered specific instructions into the Costco AI developer website and spent time training the AI to stay completely on task and be user-ready. After the AI was proven to be user-friendly, I began to insert information regarding Costco shipping into the knowledge index database. After training the model and making sure it was well versed with commonly asked questions, I double checked to see if the AI would get off topic, and ran into issues that I fixed in the database. After I thoroughly tested if the LLM was able to handle simple questions and responses, I began inserting more info into the database, such as instructions for shippers, trucking depot numbers and locations, and various protocalls. With a larger database, errors surfaced, and I spent much time improving responses through instructions and training the AI bot both through its database and its direct chat. Finally, after I had refined the chatbot and its functionality, I had to repeat the process 8 more times for the Costco websites in the UK, France, South Korea, Japan, Taiwan, China, French Canada, and Australia, each having different depots, policies, and languages. For each country, I had to train the LLM to be able to respond in English or a region's native language, based on what the user chooses to use. After a complete draft of the chatbot was finished, I read through over 5000 tickets (submitted to the tech support team) to ensure that the LLM could handle any of the questions asked by users in the past to reduce the number of questions asked to tech support. After seeing countless examples, I further refined the chatbot's functionality to have specific responses per region rather than broad solutions. Through this entire process, I learned how this AI tool influences the shipping industry, causing clarity in protocol and reducing the work of tech support and other similar services.
