Extract Data from PDFs with Azure Document Intelligence & Power Automate - No Premium Licenses

[@RogerBefurt]

Great video. This is one of the best I have seen showing how to do this. Thank you!

[@andrewhess123]

Thank you! I love this stuff and this video took a good bit of work. Thanks for watching Roger!

[@munawarhussain1400]

❤amazing tutorial ❤️

[@andrewhess123]

Thank you Munawar !

[@CédricCHEVALIER-y1d]

Awesome 😍😍😍

[@andrewhess123]

Thank you! You too! 🥳

[@saffas1866]

if the custom model extract in table format, can you advise how the compose should be like?

[@DavidGikas]

Hi, could you please go into how to restrict public access to the service but only enable it through the powerautomate of the tenant or a specific user?

[@RecognitionNowDM]

Hi Andrew, exactly what I'm looking for and you are just a savior... I have a question about using the type of flow in Power Automate. Can we make use of the 'Desktop' flow in Power Automate for this integration instead of using the Cloud flow?

[@andrewhess123]

Most likely yes, but you could always integrate a cloud flow with the desktop flow? But I still have a ton of learning to do myself with desktop flows

[@saeedsm57]

Great! Thanks!

[@andrewhess123]

Thank you Saeed!

[@CarolLoughrey]

This is perfect for what I am trying to do! Thank you! One question/problem... I am doing this with an invoice and I need the invoice total. Since it is a dollar amount, I am getting an error when I get to the last step (Updating SharePoint file properties). The error is "Input parameter 'item/AmountDue' is required to be of type 'Number/double'. The runtime value '"$850.00"' to be converted doesn't have the expected format 'Number/double'." How can I fix this? Thanks in advance!

[@andrewhess123]

Hi Carol! My guess would be to make your AmountDue field in SharePoint a string text field instead of a number field. There is def an issue with the column type it's getting confused between a string and an integer/number/double.

[@davidphilip7859]

Awesome content. I was just exploring Azure document Intelligence two weeks ago. it have me some issues cause the PDF's data which I wanted to extract had tables that were repeated with different values in it's columns

[@andrewhess123]

Thanks David! I'll write down that for the future to try out a table. It was an option in Azure the trick would be looping through each row in Power Automate from the JSON.

[@thomasgoerlach5766]

Hi Andrew, thank you very much for the excellent video. This is exactly the kind of solution I was looking for. I have successfully trained a model in Azure. Then I wanted to programme the flow. Unfortunately, I can't find the right action when I search for "form recognizer" (I use English language in the studio). My studio version (chrome) only finds "Analyze Custom Form" (no see more option) and not the "Analyze Document for Prebuilt or Custom models (v3.x API)". When I implement this, I get an error message that the model ID is wrong, though I have copied definitely the right one. I assume the reason is the wrong action. How can I find the right action? Is it called different today?

[@andrewhess123]

It was really really hidden for me too. Try clicking on the premium and filter then search for "analyze document". Then look for that symbol for Form Recognizer, its like a sheet with a text bubble coming out. I was able to find it currently today. Form Recognizer. It was buried though!

[@thomasgoerlach5766]

@@andrewhess123 Hi, many thanks. Analyze document was not successful, but analyze was working😄. Now the flow is working and I can continue to the next steps.

[@andrewhess123]

@@thomasgoerlach5766 Ha! Yeah they hid it good. Nice glad you got it working!

[@Sushant-d5r]

@@andrewhess123 i am also not able to find it any help?

[@kawaisenpai8209]

hi. This is a great video, however, can you teach me how to do for table? i have created a table in document intelligence. In my sharepoint folder, for 1 item, can it have many rows?

[@andrewhess123]

Hi Kawai, I can write this down for a future video.

[@o12jordan]

so this is just a fancy ocr document reader but also more expensive ?

[@andrewhess123]

So more expensive than what?

[@jlxe5324]

can I use it dynamically for different pdf format?

[@andrewhess123]

Sure, if you train the model. Probably use classification first.

[@andrewhess123]

This use case might be good for trying out SharePoint premium.

[@SebTheChosen]

I tried the free tier but was not able to get information from second page in document. Am i doing something wrong?

[@kcd1757]

Second page or third? The free tier is limited to the first 2 pages of a document.