ChatGPT driving me insane: it cannot retrieve the right information

[BlackCubeLabs]

Hi,

I am building this automation that takes the content of a document and then adds and updates a row on a Google sheet. The idea here is that whenever I run an interview or an AMA, I get all the content ready for another automation to start, which will create a blog post.

I am struggling to extract the exact information of the text from ChatGPT.

Here is how it is supposed to work:

• When a document is modified in a specific folder
• That document is retrieved
• I use an aggregator so that I can have the entire body of the content in one bundle
• If within the document there is a sentence like: :STATUS: INITIATE BLOG POST", then the flow continues
• ChatGPT here is instructed to get 90% of the data

ROLE DEVELOPER / SYSTEM

### OUTPUT FORMAT
The full output should be provided within the "feedback" quotes, EXACTLY in the following JSON format:

{
"email": "",
"name": "",
"blog_type": "",
"target_audience": "",
"tone_of_voice": "",
"personal_or_brand": "",
"title": "", 
"description": "", 
"blog_keywords": "", 
"blog_count_question": "", 
"Created at": ""
}


### INSTRUCTIONS
- You need to identify where each field is (e.g. "email", "name", and so on) in the body of the document and the answer given for each field
- The fields "description" might contain long text. You need to make sure to report all of it.
- Just write the output in the given JSON format and no other sentences before or after

ROLE USER

• I am just passing the array {{6.array[].text}}

• The data is then passed to a JSON parse for better classification

• A new row is added in the Google sheet - so far so good

• I have now another ChatGPT that is specifically looking to one field, called “extra_questions”; normally it is here where 99% of the content is (meaning the interview or the AMA)

ROLE USER
{{split(6.array[].text; "extra_questions")}}

ROLE DEVELOPER / SYSTEM

### OUTPUT FORMAT
The full output should be provided within the "feedback" quotes, EXACTLY in the following JSON format:

{
"extra_questions": ""
}

### INSTRUCTIONS
- You need to identify where the field "extra_questions" is in the body of the document and the answer given for that
- The field might contain long text. You need to make sure to extract all of it as it as
- You will not rewrite or paraphrase the content, you will simply extract it its integrity
- Just write the output in the given JSON format and no other sentences before or after

• Also here using JSON parse
• Then merging by updating a specific column of that new row on Google sheet

The problem is that the field with long text is not reported in its entirety. Never.
I tried to use 4o-mini, 4o, 4.1, 4.1 mini. No luck.

How can I retrieve the exact info, the full text? Perhaps I don’t even need to use ChatGPT for that…

And how can I optimize this flow? I am not even sure using an aggregator and a JSON parse are the right thing to do.

Blueprint attached.
Thanks!!

From AMA to Blog Post Automation_blueprint.json (118.4 KB)

[samliew]

@BlackCubeLabs

The problem is that the field with long text is not reported in its entirety. Never.
I tried to use 4o-mini, 4o, 4.1, 4.1 mini. No luck.

How can I retrieve the exact info, the full text? Perhaps I don’t even need to use ChatGPT for that…

How does the output of the Parse JSON modules look like?

You have not yet provided sufficient information to demonstrate the problem that would allow us to reproduce the scenario and any error(s).

To allow others to assist you with your scenario, please provide the following:

1. Relevant Screenshots

Could you please share screenshots of your full scenario? Also include screenshots of any error messages, module settings (fields), relevant filter settings (conditions), and module output bundles. We need to see what you’re working with to give you the best advice.

You can upload images here using the Upload icon in the text editor:

We would appreciate it if you could upload screenshots here instead of linking to them outside of the forum. This allows us to zoom in on the image when clicked, and prevent tracking cookies from third-party websites.

3. Output Bundles of Modules

Please provide the output bundles of each of the relevant modules by running the scenario (you can also get this without re-running your scenario from the History tab).

Click on the white speech bubbles on the top-right of each module and select “Download input/output bundles”.

A. Upload as a Text File

Save each bundle contents in a plain text editor (without formatting) as a bundle.txt file.

You can upload the file here by clicking on this button:

B. Insert as Formatted Code Block

If you are unable to upload files on this forum, alternatively you can paste the formatted bundles.
These are the two ways to format text so that it won’t be modified by the forum:

• Method 1: Type code block manually

  Add three backticks ``` before and after the content/bundle, like this:

  ```
  <sup>content goes here</sup>
  ```

• Method 2. Highlight and click the format button in the editor

  

Providing the input/output bundles will allow others to replicate what is going on in the scenario, especially if there are complex data structures (nested arrays and collections) or if external services are involved, and help you with mapping the raw property names from collections.

Sharing these details will make it easier for others to assist you.

[BlackCubeLabs]

Sure happy to provide more info. The second GPT is not picking up the content. See

[
    {
        "result": "{\n\"extra_questions\": \"\"\n}",
        "id": "chatcmpl-BeMPJ481e117EIk9D1VpqiOUsRcqo",
        "object": "chat.completion",
        "created": "2025-06-03T13:55:53.000Z",
        "model": "chatgpt-4o-latest",
        "choices": [
            {
                "index": 0,
                "message": {
                    "role": "assistant",
                    "content": "{\n\"extra_questions\": \"\"\n}",
                    "refusal": null,
                    "annotations": []
                },
                "logprobs": null,
                "finish_reason": "stop"
            }
        ],
        "usage": {
            "prompt_tokens": 3892,
            "completion_tokens": 7,
            "total_tokens": 3899,
            "prompt_tokens_details": {
                "cached_tokens": 0,
                "audio_tokens": 0
            },
            "completion_tokens_details": {
                "reasoning_tokens": 0,
                "audio_tokens": 0,
                "accepted_prediction_tokens": 0,
                "rejected_prediction_tokens": 0
            }
        },
        "service_tier": "default",
        "system_fingerprint": "fp_afccf7958a"
    }
]

[BlackCubeLabs]

I have made some steps forward. I am now able to extract some precise text using this formula:

{{slice(6.array; indexOf(map(6.array; "extra_questions"); "extra_questions") + 1; length(6.array))}}

The problem is that it does not retrieve the entire text. So perhaps I don’t even need to use ChatGPT fo this. How can I extract from the array aggregator only the text AFTER “extra questions:”?

See: