Skip vectorization and use the full document

[Azzeo]

It would be great if there was an option for CC to directly use an uploaded document (and basically add the whole document as-is in the prompt). With the advent of models allowing for a bigger context size, a RAG solution is not always the best solution for every problem anymore (for example, finding a needle in a haystack problem, or directly translating or quoting within the document as well as being less reliant on how good the embeddings model is). The current solution, by copy-pasting entire documents is not very clean and user-unfriendly.

I see two possible solutions (feel free to correct me)

1. When uploading, give the option to vectorise it, or directly add it to the prompt
2. When the user intent is to perform a specific task for an entire document or group of documents that were previously uploaded, the entire raw content of the document(s) should be added to the prompt. I am not aware if SK offers any of this functionality, but it would be great to see to see some kind of solution for this.