Efficiency puzzle for complicated handwriting to notion workflow

Questions
, , , ,
1

I recommend only looking at this if you like a challenge!

I’ve now got a workflow running that takes handwritten pdf files from an e-ink tablet and converts to text entries in Notion. Because I was new to Make and low-code in general I suspect there are smarter, more efficient ways to do this with less operations. So consider this a challenge if you know what you’re doing with Make, how could this be made more efficient? Creative ideas very welcome, I’m here to learn :slight_smile:

Blue print

Handwriting to Notion blueprint.json (590.9 KB)

Visual

image
image3615×1578 451 KB

Workflow Overview

  1. Webhook Trigger:
  • The workflow starts when a file is uploaded or changed in Dropbox, triggering the webhook.
  1. Fetch Metadata:
  • It retrieves metadata (like the file path and cursor) from Dropbox to identify new or modified files.
  1. Filter Files:
  • Only specific files are processed, filtering by file type and excluding certain folders.
  1. Generate Temporary Links:
  • Temporary download links are created for files in Dropbox to access their content.
  1. PDF Conversion:
  • If the file is a PDF, it is converted into images using an external API (e.g., PDF.co).
  1. OCR with OpenAI:
  • The images are sent to OpenAI for OCR (Optical Character Recognition), extracting text from the handwritten or printed content.
  1. Process and Format:
  • The extracted text is structured into Markdown for better organization and usability.
  1. Search or Create Notion Entries:
  • The workflow checks if a corresponding entry already exists in the Notion database using the file’s metadata.
  • If no entry exists, a new page is created with the processed data.
  1. Update Notion Database:
  • The workflow updates the database fields, including tags, extracted text, and any relevant metadata from Dropbox.