Gold Retriever is an open-source tool that allows ChatGPT, a language model, to store, retrieve, and reason with user-specific data. By creating a custom plugin through Gold Retriever, users can enable ChatGPT to seamlessly integrate with and reason over new, personalized or up-to-date data, overcoming ChatGPT’s inherent limitations such as its knowledge cut-off point or struggle with real-time data. Gold Retriever acts as a bridge linking ChatGPT with the vast and dynamic world of data, making AI interactions more meaningful, accurate, and personal. To utilize Gold Retriever, users can gather relevant text files in one directory and install Gold Retriever via pip. Then, users can deploy a plugin that provides the OpenAI key, index the data, and integrate the plugin into ChatGPT. The case study showcases how students can use Gold Retriever to assist with coursework, creating flashcards from course materials and utilizing ChatGPT’s help for this task. However, there are numerous applications for Gold Retriever beyond this, as long as there is textual data that users would like ChatGPT to remember and use. Overall, Gold Retriever allows ChatGPT to interact with user-specific data, enabling the model to become more personalized and accurate in its outputs.
