[422] error when trying to insert GPT-generated tags into a multi-select field in Airtable using Make

Hi all — I’m encountering a persistent [422] error when trying to insert GPT-generated tags into a multi-select field in Airtable using Make

Scenario Overview

  • I’m using a GPT (OpenAI) response that returns a string of tags like:

mathematica

CopyEdit

Part A, Part L, Structural
  • I split this string into an array with:

make

CopyEdit

split(ComplianceTagsRaw; ", ")

(stored in a Set Variable called ComplianceTagsArray)

  • I then try to insert this into Airtable’s Compliance Tags field, which is:
    • A multi-select field
    • With all expected tags predefined (Part A, Part L, etc.)
    • Map ON is enabled

:red_exclamation_mark: Problem

Despite this, Airtable keeps returning:

pgsql

CopyEdit

[422] Insufficient permissions to create new select option ""split(ComplianceTagsRaw; \", \")""

Or:

csharp

CopyEdit

[422] Cannot parse value for field Compliance Tags

I’ve also tried using a Text Aggregator, but it results in this error:

bash

CopyEdit

[422] Insufficient permissions to create new select option ""{\"text\":\"{{3[7][1]}}\"}{{3[7][1]}}""

This tells me Make might still be passing a raw string or an object when Airtable wants a flat string array.


:white_check_mark: What I’ve Confirmed So Far

  • ComplianceTagsArray is an actual array: ["Part A", "Part L", "Structural"]
  • Predefined options in Airtable match GPT output exactly
  • Map ON is toggled correctly
  • All variables are resolved using {{ }} and not typed manually
  • Wand icon is used to select values (not pasted formulas)

:puzzle_piece: Question

How can I reliably insert a split GPT tag list into a multi-select field in Airtable using Make?

  • Do I need to use an array aggregator with a specific structure?
  • Should I format it as an array of text objects or just strings?
  • Is there a known quirk where Make sends an unresolved formula when mapping from a Set Variable?

Any screenshots, JSON example formats, or full working flows would be hugely appreciated.

Thanks in advance!