Hmmm…

My inclination would be to extract the keywords from Airtable and inject them into your ChatGPT prompt, instructing ChatGPT to ignore those keywords.

Another option is to have an action within your ChatGPT assistant that calls Airtable.

L

Thanks! I attempted that, but technically, I don't want to disregard them. My goal is to utilize them first, tag them if they meet the criteria, and then create new ones if they don't.

You can instruct callin.io to send all the keywords to ChatGPT within the prompt and direct ChatGPT to prioritize using those keywords.

I believe this is more of a prompt engineering challenge. You might consider structuring your prompt as follows:

#TASK

Create a list of new keywords extracted from the content of a website page. Only include keywords that are not present in the existing keyword list.

#KeywordList

• Populate this with all your keywords from Airtable

#WebsiteContent

• Insert the website content here

I would recommend testing this manually within ChatGPT first until you refine the prompt to achieve the desired outcome. Once it functions correctly, you can then integrate it into your callin.io automation.

L

P.S. I have a tool that utilizes ChatGPT for this purpose. I input your information, and this is the output I received:

Role (R):

You are an assistant responsible for parsing keywords from website content and comparing them against a pre-existing list of keywords from Airtable.

Objective (O):

Ensure that keywords already present in Airtable are utilized first, marking them as "matched." Generate new keywords only when no close matches are found in Airtable.

Context (C):

1. The user is extracting keywords from a website.
2. The user maintains an existing keyword list in Airtable, which includes variations (e.g., "Department" versus "department").
3. You will be provided with a list of keywords from Airtable to check against before creating new ones.

Procedure (P):

1. Analyze the provided text and identify potential keywords.
2. Compare each identified keyword with the list supplied by the user from Airtable.
3. If an exact match or a close match (disregarding case and minor spelling differences) is found, tag it as "matched."
4. If no match is identified, create and return new, non-duplicated keywords.
5. Present a final list comprising both matched and newly generated keywords.

Output (A):

Display the results in two distinct sections:

1. Matched Keywords (with their respective tags)
2. Newly Generated Keywords

Here's a suggestion:

1. Run GPT to generate the keywords.
2. Combine the keywords from the Airtable list and the generated ones into a single text string, separated by a delimiter such as a comma.
3. Utilize the lower() function to convert all keywords to lowercase.
4. Employ the split() function to transform the keywords into an array.
5. Use the deduplicate() function to eliminate any duplicate entries.
6. Subsequently, you can rejoin them into a string if desired.

Hope this assists!