Answered

Getting error while using AI builder to convert pdf file to excel in flow

(0) Share

Report

Posted on by KanwalKaurSaini

Hi All,

I am trying to use AI Builder custom form processing model to convert pdf files received in my sharepoint folder to excel but i Am getting below error:-

"For application/pdf mime-type The input file appears to be corrupted and cannot be processed.; transactionId=eLRLagBmRqgRwTApAuCGOBg233PDuGYs-PowerAutomate".

The design of the flow is as below :-

Please help me how should I remove the error .

My actual requirement is convert the pdf files received to excel and store in different folder. Is it possible to convert selective pages of pdf file instead of whole pdf file in any way . I tried using AI builder but it gave me above mentioned error.

if anyone can help me with achieving my usecase.

Thanks

Kanwal Saini

Categories:

AI Builder

I have the same question (0)

All responses (5)

Answers (1)

Verified answer

JoeF-MSFT on at

Like (0)

Report

Hello @KanwalKaurSaini,

Thanks for the question. It looks like you are building a great automation!

The AI Builder Form Processing action will extract the content from the PDF. What you can then do is take that extracted data and use the Excel connector to insert directly it into an Excel file like this:

Note that the Excel file will need to be in folder in SharePoint or in OneDrive, and will need to have a table defined with the columns you need.

I hope this helps! Let us know if you have further questions.

Was this reply helpful? Yes No
KanwalKaurSaini 43 on at

Like (0)

Report

Thanks , I will try this out .
Is it possible to select particular pages of pdf specific to first 2 pages only to convert them to excel and then store in different sharepoint folder . Also Can we send different pages as different tabs in excel ?

Was this reply helpful? Yes No
JoeF-MSFT on at

Like (0)

Report
Hi @KanwalKaurSaini,

In a few weeks we will introduce the option in the AI Builder action in Power Automate to specify which pages you want to process. But as of today, all pages from the document are processed by the AI model. Something you can do is:

Define in one tab of your Excel file one table, and in a second tab another table. You define tables in Excel like this: Create a table in Excel - Excel (microsoft.com)

In your AI Builder Form Processing model you defines two tables like this:

You then tag each table accordingly in your documents.

Once you have trained and published your model, you use it in a flow and save the outputs of the table page 1 in the first table of the Excel file, and the outputs of table page 2 in the second table of the Excel file. This page explains how to handle tables from the AI model in a flow: https://docs.microsoft.com/en-us/ai-builder/form-processing-model-in-flow#iterate-a-form-processing-table-output-in-power-automate

Keep us posted on the progress you make in this automation.

Was this reply helpful? Yes No
KanwalKaurSaini 43 on at

Like (0)

Report

Hi ,

Thanks for your solution.IT worked fine .
I created 2 sheets in excel with required columns and then as per your solution, i used 2 parallel add new rows action in my flow .It worked . Now the requirement is modified . I also want to showcase the page no also from where the data is taken into my excel as in actual case my PDF file would be having 200 pages and my AI builder should process selective pages into my excel sheet and then using that to show in power bi report. to fulfill this requiremennt , i created 4 paged pdf and performed testing with Ai builder . Sharing my observations below :-
1) in my first test, i trained my model to detect customer name from page 1 and item table from page 2 but the Ai builder processed both information from page 1 only .
2) in my 2nd test , i vice versa the trained model in step 1 .then the Ai builder processes information from both pages into my excel sheet.
3) in my 3rd and 4th test , i trained my model to detect info from page 2 only then Ai builder processed page 2 only into excel.
4) in detecting multiple pages , Ai builder capacity is also used a lot . in my case as an example it was previously 43 pcnt remaining now it is only 37 pcnt .
how can we show page no also.how can the performance of Ai builder can be improved in my requirement. is it feasible approach ?

Was this reply helpful? Yes No
KanwalKaurSaini 43 on at

Like (0)

Report

Hi all,
I created AI model for paged invoice to check the feasibility and business value in conversion of PDF invoicefiles data to excel . THe behaviour of trained model is not constant every time the flow runs . It extracts different data into excel not from the pages I trained. Adding my test analysis for others 's reference . Morover it uses lot of credits in analysing bigger file every time. So We concluded from the test its not feasible with AI builder to convert particular pdf page to excel and advised our client to use other solutions. If any leads have achieved it with power platform please do let me know.

THanks for wonderful ideas so far 🙂

AI Builder test r...

Was this reply helpful? Yes No