Step 2: Set Up Topic Inputs

Map the topic details that the AI will use to generate the outline and subsections.

1. Open Topic Input Mapper.
2. Set Topic to Input your topic Here.
3. Set How many outline? to 3.
4. Set Language to Input the language here (ex: English/Indonesia/Arabic/etc).
5. Set Keyword to Put your keyword here.

Step 3: Configure the Outline Generation AI

This stage uses AI to generate a structured outline and validates the JSON output format.

1. Open AI Chat Engine and select your model (configured as gpt-4o-mini).
2. Credential Required: Connect your openAiApi credentials in AI Chat Engine.
3. Open Outline Builder and set Text to =Write a creative outline max {{ $json["How many outline?"] }} item on {{ $json.Topic }} in {{ $json.Language }} language. Output a JSON object with the following structure: { "topic": "string", "subtopics": [ { "subtopic": "string", "details": "string" } ] } Start the subtopic with an Introduction and end with a Conclusion..
4. Ensure Outline Builder has Has Output Parser enabled and is connected to Structured Result Parser.
5. In Structured Result Parser, set JSON Schema Example to { "topic": "string", "subtopics": [ { "subtopic": "string", "details": "string" } ] }.

Step 4: Extract Subtopics and Generate Section Drafts

Convert the outline into subtopic items, loop through them in batches, and draft a section for each one.

1. In Extract Subtopic List, set JavaScript Code to // Loop over input items and add a new field called 'myNewField' to the JSON of each one return $input.first().json.output.subtopics.
2. Connect Outline Builder → Extract Subtopic List → Iterate Items Batch.
3. Connect Iterate Items Batch to Content Draft Agent for drafting each subtopic section.
4. In Content Draft Agent, set Text to =The outline refers to parts of a conversation. This is the article section that covers these topics: Subtopic: {{ $json.subtopic }} Details: {{ $json.details }} Write a maximum of 1 paragraph. Do not number any sections. Include a lot of detail. Mandatory Rules 1. Do not add numbering in any part. 2. Write with detail and dense information. 3. Ensure more than 75 percent of sentences contain a maximum of 20 words to meet YOAST SEO readability standards. 4. Ensure the keyword "{{ $('Topic Input Mapper').item.json.Keyword }}" appears in the sentences. 5. Use clear and active sentences. 6. Use 1 or 2 transition words recognized by YOAST. Examples: so, therefore, however, besides that, then, after that, for example. 7. Avoid emojis, excessive punctuation, and promotional style. 8. Do not use em dashes to keep the writing natural and not AI generated. 9. Write in {{ $('Topic Input Mapper').item.json.Language }} Language.
5. Open Secondary Chat Engine (model gpt-4o-mini) and ensure it is connected as the language model for Content Draft Agent.
6. Credential Required: Connect your openAiApi credentials in Secondary Chat Engine.
7. In Window Memory Buffer, set Session Key to ={{ $('Iterate Items Batch').item.json.details }} and Session ID Type to customKey (credentials are handled by the parent Content Draft Agent).
8. After drafting, connect Content Draft Agent → Append Subtopic Field → Iterate Items Batch to loop through the next subtopic.

Step 5: Aggregate Drafts and Build Final HTML

Combine all drafted sections into a single output and format it for publishing.

1. Connect Iterate Items Batch → Combine All Items to aggregate all item data.
2. In Combine All Items, set Aggregate to aggregateAllItemData.
3. Connect Combine All Items → Build HTML from JSON.
4. In Build HTML from JSON, keep the provided JavaScript Code that reads $('Outline Builder').first().json.output.topic and compiles content_markdown and content_html.

Step 6: Test and Activate Your Workflow

Validate each stage and confirm the output structure before using it in production.

1. Click Execute Workflow to run Manual Launch Trigger manually.
2. Verify Outline Builder returns a JSON object matching the schema in Structured Result Parser.
3. Check Combine All Items for a single aggregated array and confirm Build HTML from JSON outputs title, content_markdown, and content_html.
4. When results look correct, toggle the workflow to Active for production use.