From API to AI: How We Taught an LLM to Speak Canvas
Think of a Large Language Model as a brilliant, creative mind. Now, think of a web API like Canvas as a giant, old-school filing cabinet. The brilliant mind can't just pull open the drawers; it needs an expert assistant who knows the filing system. This paper is the story of how we built that expert assistant.
Think of a Large Language Model as a brilliant, creative mind. Now, think of a web API like Canvas as a giant, old-school filing cabinet. The brilliant mind can't just pull open the drawers; it needs an expert assistant who knows the filing system. This paper is the story of how we built that expert assistant. We'll walk you through the big problems—like APIs being slow and demanding cryptic codes instead of plain English—and how we solved them by building a smart, local 'brain' for the AI to use. We'll also introduce our autonomous 'librarian' software that keeps this brain stocked with the latest information, and most importantly, how we taught the system to use common sense to answer questions like 'Where's my math homework?'. This is the technical story of building a bridge between the world of AI and the world of APIs.