Does anyone have interest in something like

tap-chatgpt-vector-data

/

tap-chatgpt-training-data

whiich would take documentation refs or other input and generate vectors and supporting metadata for training ChatGPT applications? Interest in using and/or helping build it? As I've been researching how we might get started with a chatbot for Meltano docs, it seems the first big challenge everyone faces is how to create vectors from the raw markdown or text inputs. Maybe I'm oversimplifying here, but it seems the "tap" paradigm would be a great one for this use case. Downstream processing and service scaffolding would not be included in this step, although in theory a future Meltano utility might be able to help with those next steps. More info in 🧵.