Skip to main content

Hi everyone, 

I’m trying to create a p2p library app. 

I’d the user experience to be:

  • User sends a photo of their bookshelf to an email address
  • They see their books (and those of other people) in an airtable, where each row is a book (with title, author, and owner)

The current approach I’m trying:

  1. Trigger - New Email Matching Search in Gmail (working)

  2. Action - OCR With File in Nanonets OCR (working - see attachment for test output)

  3. Action - Extract Structured Data in ChatGPT (not working - see attachment for test output)

  4. Then not sure how to go from a list of books, into an airtable with one row per book]

 

I’d really appreciate any advice and help on:

  1. Is this the optimal overall approach?
  2. How can I get ChatGPT working? Am I using the fields wrong?
  3. Advice on how to go from a list from Chatgpt, to the airtable? So far I am considering using text formatter + “create multiple spreadsheet rows” from Google Sheets, and then going to airtable from this thread: 

 

Hi @Stru Delman 

Good question.

You can use the Looping app to iterate over each item: https://zapier.com/apps/looping/help

You can use the Formatter > Utilities > options to prepare the line items. (e.g. Text to Line Items)


Hi @Troy Tessalone , 

Thanks for your kind reply. I haven’t been able to get it working that way, but I think I have reframed the challenge to be more straightforward. 

I have this output from ChatGPT: 

(ChatGPT can reformat with commas or spaces etc however is best).

And want it to end up in a Google sheet with this format:
 


The challenge is I want to separate this block of text into both rows and columns, and I’m not sure how to go about it.

Any advice on how to do this would be much appreciated. Thank you for any help in advance!​​​​​​​


@Stru Delman 

You may want to play around with this Zap action: ChatGPT - Extract Structured Data

You may need to instruct AI to delimit the data with pipes || instead of commas to make it easier to parse.

Then you can instruct ChatGPT to create an array from all the values before the first pipes as the Book Title, etc.

 

Or a Code step can be used.

 


Thanks for the inputs @Troy Tessalone . 

In the end, I’ve got it working by using the text formatter, following this approach: 


I couldn’t figure out the approaches you hinted at, but I will keep trying, because I think they may be more efficient (fewer zaps).

Thanks for your help!​​​​​​​


That’s fantastic news, @Stru Delman🎉

Thank you so much for following up here to share how you solved this. It’s most appreciated! 🤗