My workflow needs to identify keywords within emails. However, the matching process cannot be exact, as emails might contain various forms of the keywords.
I considered using AI for this. Based on my limited testing, it appears to be effective.
Previously, my workflow made an API call for each keyword listed in a Google Sheet. This approach consumed a significant number of operations and API credits.
Therefore, I decided to consolidate the keywords into a JSON string and execute a single prompt against this JSON. So far, this method seems to be working.
The JSON structure is quite simple, with each item containing two keys: an email address and a keyword.
The prompt's function is to check if any variation of the keyword is present in the email. If a match is found, it forwards the email to the associated email address.
Do you believe this approach is reliable?
Here is the blueprint:
blueprint (8).json (90.4 KB)