Step 2: Connect Scrapeless and Parse Blog Content

This branch crawls blog pages, extracts links and content, and batches the results for aggregation.

1. In Crawl Blog Pages, set URL to https://www.scrapeless.com/en/blog, Resource to crawler, Operation to crawl, and Limit Crawl Pages to 20.
2. Credential Required: Connect your scrapelessApi credentials in Crawl Blog Pages.
3. In Parse Content Details, keep the provided JavaScript to extract title, mainContent, and extractedLinks.
4. In Split Links and Text, set Field to Split Out to extractedLinks.
5. In Scrape Full Content, set URL to {{ $json.url }} and Resource to crawler.
6. Credential Required: Connect your scrapelessApi credentials in Scrape Full Content.
7. Keep Iterate Batch Items connected so it loops through scraped pages and passes batches to Aggregate Records.

Step 3: Aggregate, Convert, and Store Content in Pinecone

This step aggregates scraped content into a file, chunks it, and inserts embeddings into Pinecone.

1. In Aggregate Records, set Include to specifiedFields, Aggregate to aggregateAllItemData, and Fields to Include to markdown.
2. In Convert Data to File, set Operation to toText and Source Property to data.
3. Configure Recursive Text Chunker with Chunk Size 2000 and Chunk Overlap 200.
4. In Standard Data Loader, set Data Type to binary to load the file output.
5. In Pinecone Index Writer, set Mode to insert, Pinecone Namespace to DataPlace, and select Pinecone Index seo-writer.
6. Credential Required: Connect your pineconeApi credentials in Pinecone Index Writer.
7. Gemini Embedding Builder is connected as the embedding model for Pinecone Index Writer — ensure credentials are added to the parent connection for the AI embedding input (not the sub-node).

Step 4: Configure Keyword Planning and Content Formatting

This branch builds keyword strategy from SERP data, then formats the AI output into HTML.

1. In Assign Keyword Fields, set Keywords to "Scraping", "Google trends" and Search Intent to People searching to get tips on Scraping.
2. In Analyze SERP Keywords, set q to {{ $json.Keywords }}.
3. Credential Required: Connect your scrapelessApi credentials in Analyze SERP Keywords.
4. In LLM Keyword Planner, keep the Text prompt as provided, including the expressions {{ $('Assign Keyword Fields').item.json.Keywords }}, {{ $('Assign Keyword Fields').item.json['Search Intent'] }}, and {{ JSON.stringify($json.organic_results) }}.
5. Gemini Chat Model A is connected as the language model for LLM Keyword Planner — ensure credentials are added to the parent connection for the AI language model input (not the sub-node).
6. In Markdown Converter, set Mode to markdownToHtml and Markdown to {{ $json.text }}.
7. In HTML Formatter, keep the HTML template and ensure the content placeholder uses {{ $json.data }}.

Step 5: Set Up Conversational Retrieval and AI Reasoning

This branch enables chat-based retrieval from Pinecone and uses a reasoning agent to generate answers.

1. Use Chat Message Trigger as the entry point for chat-based queries.
2. In Pinecone Index Reader, set Mode to load and Prompt to {{ $json.chatInput }}, with Pinecone Namespace DataPlace and Pinecone Index seo-writer.
3. Credential Required: Connect your pineconeApi credentials in Pinecone Index Reader.
4. In AI Reasoning Agent, set Text to {{ $('Chat Message Trigger').first().json.chatInput }} and keep the system message template for contextual answers.
5. In Conversation Buffer Memory, set Session Key to {{ $('Chat Message Trigger').first().json.sessionId }} and Session ID Type to customKey.
6. Gemini Embedding Engine is connected as the embedding model for Pinecone Index Reader — ensure credentials are added to the parent connection for the AI embedding input (not the sub-node).
7. Gemini Chat Model is connected as the language model for AI Reasoning Agent — ensure credentials are added to the parent connection for the AI language model input (not the sub-node).

Step 6: Test & Activate Your Workflow

Run a full test of both branches to verify data flow and AI outputs before enabling production use.

1. Click Execute Workflow from Manual Launch Trigger and confirm that HTML Formatter produces a formatted report.
2. Send a test chat message through Chat Message Trigger and verify AI Reasoning Agent returns a response using Pinecone context.
3. Check the execution log to confirm Pinecone Index Writer inserted vectors and Pinecone Index Reader returned documents.
4. When successful, toggle the workflow to Active to enable chat-based retrieval in production.