One super useful and often overlooked feature of vega lite charts is they embed the data within the chart definition. That means you can easily extract the raw

One super useful and often overlooked feature of vega lite charts is they embed the data within the chart definition. That means you can easily extract the raw data. You can also easily change the visual appearance or chart type by opening the chart definition in the vega lite editor. https://vega.github.io/editor/ https://vega.github.io/editor/ Another aspect in this new world of LLMs the since they're declaratory, you can just get the LLM to write or edit the spec ('give the axes sensible titles and ensure the colours are accessible'). I've found in practice even gpt4 only gets it right for relatively simple charts, but I'm sure this will improve