How to watch a folder for a text file, summarize it with ChatGPT, and create a file in Google Drive Obsidian Vault?

Show & Tell
1

This topic was automatically generated from Slack. You can find the original thread here.

Similar to the one above but for text instead of audio. Watch a folder for a text file, summarize it with ChatGPT, then create a file in a Google Drive Obsidian Vault.
Trigger: New file in a Google Drive folder

  1. Download file
  2. Bash script cat the file to `PIPEDREAM exports.
DATA=`cat /tmp/text.txt`

EXPORT="contents=${DATA}"
echo $EXPORT >> $PIPEDREAM_EXPORTS
  1. ChatGPT to summarize
  2. Create file from text in Google Drive

image.png
image.png1838×1446 197 KB

image.png
image.png1828×1558 220 KB

image.png
image.png1848×906 41.5 KB

image.png
image.png1832×1808 320 KB

image.png
image.png1852×1828 349 KB

image.png
image.png1836×1602 255 KB

1 Like
2 
Write a Title for the transcript that is under 15 words and contains no quotation marks

Then write: "---"

Write "Summary" as a Heading 2 (For example: ## Heading)

Write a summary of the provided transcript.

Then write: "---".

Then return a list of the main points in the provided transcript. Then return a list of action items. Then return a list of follow up questions. Then return a list of potential arguments against the transcript.

For each list, return a Heading 2 (For example: ## Heading) before writing the list items. Limit each list item to 100 words, and return no more than 5 points per list.

Transcript:
{{steps.code.contents}}

And:

Write a Title for the transcript that is under 5 words, containing no quotation marks or special characters

Transcript:
{{steps.chat.$return_value.original_messages_with_assistant_response[1].content}}
3

hi - I’m trying this but having issues with the ChatGPT step timing out despite set at 180s / 4gb memory, even testing with a very short text file. any ideas?

4

hmm, i’m not sure! Perhaps the API itself was having some issues?

5

I was trying to do the exact same thing.

  1. google drive trigger to watch for text file (success)
  2. download the file to tmp/ (success)
    3 use chat gpt to summarize ( pass with warning.)
    Upon closer inspection it looks like chat gpt is not looking at the contents of the file. just the file itself so it’s not summarizing anything. Do you think you have the same issue?