Any idea why Argon will randomly add additional context to answers? Created a column w/ Use AI (Argon) to extract a zip code from an address. The majority worked correctly, but then others gave unnecessary reasoning and details.
Yep - that’s normal when the AI isn’t tightly constrained, so some rows “helpfully” add reasoning or extra text instead of returning only the ZIP. [1]
To make it consistently output just the ZIP:
Update the prompt to explicitly require a single value (e.g., “Return ONLY the 5-digit ZIP code. No words, no punctuation, no explanation.”). [2]
If possible, constrain the output to a structured format (JSON schema with a zip field) so it can’t free-form.
Use Preview and iterate on the prompt until every preview row returns the same format. [3]
Also worth double-checking your column configuration matches across runs, since differing AI options can change output behavior.
Did that answer your question?
If you still need help refining the AI output to consistently return only the ZIP code, I’m here to assist. Would you like to provide more details about what you’re trying to achieve or any specific challenges you’re facing?
what is your prompt
I don’t have your exact prompt, but here’s a solid one that usually prevents Argon from adding extra explanation by forcing a strict output format. [1]
#VARIABLES#
{Address} = /Address
#INSTRUCTIONS#
Extract the 5-digit US ZIP code from {Address}.
Return ONLY the ZIP code (5 digits).
No words, no punctuation, no reasoning.
If no ZIP code is present, return blank.Did that answer your question?
If you still need help with refining your prompt or AI output, I’m here to assist. Would you like to share more about what you’re trying to achieve or any specific issues you’re encountering?
