On making charts extractable - that's a good point. I haven't thought about it so far. Curious about what you'd want here: the chart as an image, an embed, or the underlying data as a CSV?

On RAG - not quite. All datapoints are fetched from trusted sources via API or web search. An LLM interprets the question and then routes it to a source. I am not sure if RAG could fetch live datapoints.