Spreadsheets as Knowledge Base Documents

A basic question, I know, but I’ve looked and I can’t seem to find an answer in any of the previous posts.

My knowledge base is contained in a series of Excel spreadsheets. What is the best way of loading these into Pickaxe? As Excel files, as CSVs, or as PDFs?

Hey @battleguide you can upload both PDF and CSV into your Pickaxe’ KB. For spreadsheets, CSV file format is the best.

A great way to find out if your spreadsheets is being comprehensively ingested by the KB is to get a count of the data cells in the sheet and then comparing that against the # of ‘Chunks’ listed next to your CSV file info (after uploading it to your KB.

Thank you for the quick response Ned. And also for your very informative videos. I’m not certain I fully understand your second paragraph - but I’m sure I will eventually :wink::smiling_face_with_sunglasses:

1 Like

@battleguide for a detailed look at how the KB works (and interacts with spreadsheets!), you can also check out this blog post!

This blog post is incredibly useful. Thank you.

@battleguide No problem :wink:

I’ll explain the kb doc ingestion process.

Whenever you upload a file to your Pickaxe’ knowledgebase the data gets ingested and processed so that it can be scanned and retrieved to answer user queries.

When you upload. cSV files they will be broken down into data chunks that are mentioned and measured next to each file you upload to the Pickaxe KB.

When you’re done uploading a file to your kb, wait until it’s processed and check the # of chunks that were ingested (next to the kb file name to the right). The # of chunks should be close to the # of cells present in your CSV file with a small margin difference (3-4 chunks missing for e.g.) which is normal. If the # of chunks varies by a vast amount, delete the CSV and then reupload it to reprocess it.

I hope that makes sense!

1 Like