I want to use an AI model to execute a checklist of rules against an excel workbook

[dob-xml]

My scenario is that I would like to use a model to analyse an excel workbook e.g. prompts may be

"How many worksheets are in the excel workbook here https://xxx.blob.core.windows.net/general/example.xlsx"
"Are all of the values in the cell range A1:D5 positive integers, if not list the cell name and the value"
"How many cells in column A have non-blank values"

How do I go about making the file available to the model for analysis?

I will be looking for an programmatic way of achieving this.

Ideally I would like to use the Agent flow in Logic Apps but I can work with any API to upload the file and pass the prompt.

[chanezon]

You could use the Microsoft Graph API to access the Excel workbook programmatically. Then send the data to a model.
https://learn.microsoft.com/en-us/graph/api/resources/excel?view=graph-rest-1.0

[dob-xml]

Thank you for that. I was hoping to avoid having to do something like that. Sending the data in parts misses my objective of finding a model that understands the structure of xlsx. Using Graph API means I would have to write code when new prompts are created. I know some models have some understanding as I have conducted some simple prompting in a chat playground.

[garrytrinder]

You can use the Open XML SDK for Office SDK (this is C# only) to access the contents of an Excel spreadsheet, however if you are looking for language alternatives then there are third party libraries available that can parse OpenXML documents.

All Office files are based on OpenXML, it is possible to change the extension to ZIP and extract the contents, however your one file is split up into multiple files. You maybe able to pass the contents of these files as context to a model to work with.

[dob-xml]

Thanks that might work