Hey ​@BuildForAT!

Congrats. Couple of questions.

1. What tool did you use for the video (looks nice!)
2. How are you thinking about the dynamic generation of fields depending on the document type that end user provides?

Best of lucks!

Mike, Consultant @ Automatic Nation 

Hey ​@Mike_AutomaticN,

Thank you! 

1. I used a range of tools. VN editor, Pexels for the stock footage, openai.fm for the voiceover.

2. I’m thinking that the structure of the table will be predefined by the user. The dynamic part will be the processing of the document data using a combination of OCR and AI tools, to figure out what data goes into each field.

I hope this clarifies things.

Do you think a tool like this would benefit any of your clients? I’m sure someone as experienced as you has encountered many kinds of Airtable users!

 

Hey ​@BuildForAT!

Did you check out Airtable’s native AI fields? Once you’ve created them you can prompt each of them individually to get one specific aspect of the pdf. Of course the prompt can make reference to the Attachment field containing the pdf, so no OCR or additional third party AI tools needed.

For sure this will not automatically push to a neatly structured schema of different tables but only to the specific record within the specific table, but some workarounds could be probably built using automations.

Mike, Consultant @ Automatic Nation

Thanks for telling me about this existing ability. I didn’t know the AI fields could do that! I thought that it could only process other text fields. I’ll definitely check it out because it could save me a bunch of time.