Step 2: Connect Google Sheets

These nodes read and write the website content database used by the assistant and by the summarization process.

1. Open Retrieve Data Schema and select the spreadsheet in Document ID and Sheet Name (e.g., gid=0).
2. In Retrieve Data Schema, ensure the filter uses Lookup Column Data schema with Lookup Value ={{ true }}.
3. Credential Required: Connect your googleSheetsOAuth2Api credentials in Retrieve Data Schema.
4. Open Schema Flag Update, set Operation to appendOrUpdate, and map Data schema to ={{true}}.
5. Credential Required: Connect your googleSheetsOAuth2Api credentials in Schema Flag Update.
6. Open Append Sheet Row, keep Operation as append, and confirm column mappings use the AI expressions like ={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Summary_Content', ``, 'string') }}.
7. Credential Required: Connect your googleSheetsOAuth2Api credentials in Append Sheet Row.

Step 3: Set Up URL Validation and Schema Checks

This section validates user URLs, checks whether the schema is already present, and routes into either the chat response or sitemap scanning flow.

1. Open Schema Check and confirm the condition uses ={{ $json['Data schema'] }} with a boolean true check.
2. Open URL Validator Agent and set Text to ={{ $node["Web Chat Trigger"].json["chatInput"] }}.
3. Attach Structured Parser to URL Validator Agent as the output parser and keep the JSON schema example:

{ "URL": "example.com", "URL_bool": true }

1. Connect OpenAI Validator Model as the language model for URL Validator Agent.
2. Credential Required: Connect your openAiApi credentials in OpenAI Validator Model (credentials are set on the model node, not the agent).
3. Open URL Valid Check and confirm the condition uses ={{ $json.output.URL_bool }} with a boolean true check.
4. For invalid URLs, ensure Invalid URL Reply sends the message Debes introducir una URL válida ejemplo: https://google.es.

Step 4: Configure the Conversational Agent and Memory

This agent answers chat requests using the website data in Google Sheets and a language model, with memory enabled for context.

1. Open Conversational Agent and set Text to ={{ $node["Web Chat Trigger"].json["chatInput"] }}.
2. Connect OpenAI Chat Engine as the language model for Conversational Agent.
3. Credential Required: Connect your openAiApi credentials in OpenAI Chat Engine.
4. Attach Lightweight Memory to Conversational Agent and keep Context Window Length at 50.
5. Attach Fetch Sheet Rows as an AI tool to Conversational Agent so it can retrieve website content from the sheet.
6. Credential Required: Connect your googleSheetsOAuth2Api credentials in Fetch Sheet Rows (tools use their own credentials, but are invoked by the agent).
7. Confirm Conversational Agent routes to Agent Response and that Agent Response uses ={{ $json.output }} for its message.

Step 5: Configure Sitemap Discovery and Selection

These nodes build a valid sitemap URL from robots.txt, fetch the sitemap, and select the correct sitemap section using AI.

1. Open Scan Options and keep the boolean flags as configured: scan_pages true, scan_posts false, category false, tags false.
2. In Random UA Picker, keep the provided JavaScript that returns a random userAgent string.
3. Open Robots.txt Request and set URL to ={{ $node["URL Validator Agent"].json["output"]["URL"] }}/robots.txt.
4. Confirm Parse Sitemap URL parses the robots.txt content and returns sitemapUrl as coded.
5. Open Sitemap Fetch and set URL to ={{ $json.sitemapUrl }} with headers including User-Agent ={{ $json.userAgent }}.
6. Open XML Decode A to parse the sitemap XML and pass it to Sitemap Selector.
7. Open Sitemap Selector, keep JSON Output enabled, and verify it references scan options and the sitemap index locs using expressions like {{ $json.sitemapindex.sitemap[0].loc }}.
8. Credential Required: Connect your openAiApi credentials in Sitemap Selector.
9. Open Page Sitemap Fetch and set URL to ={{ $json.message.content.sitemap_page }}, then route to XML Decode B.

Step 6: Configure URL Splitting, Page Requests, and Summarization

This stage expands the selected sitemap into individual URLs, fetches each page, converts HTML to Markdown, and summarizes the content with AI.

1. Open Combine URL Map and keep the JavaScript that transforms urlset.url into a urls object.
2. Open Split URL Items and set Field To Split Out to urls.
3. Open Batch Iterator to control batch processing of URLs before requesting pages.
4. Open Page Request and set URL to ={{ $('Split URL Items').item.json.urls }} with User-Agent header ={{ $json.userAgent }}.
5. Open HTML Convert to MD and set HTML to ={{ $json.data }}.
6. Open Page Summary Model and ensure the message includes =URL: {{ $('Split URL Items').item.json.urls }} {{ $json.data }} in the prompt.
7. Credential Required: Connect your openAiApi credentials in Page Summary Model.
8. Confirm Page Summary Model uses Append Sheet Row as an AI tool to write results to the sheet.

Step 7: Add Error Handling

The workflow includes explicit stops for invalid URLs and sitemap failures.

1. Verify Request Error Halt is connected from the error output of Robots.txt Request and uses the message URL mal introducida, debes introducir con el siguiente formato: ejemplo.com.
2. Verify Sitemap Failure Halt is connected from the error output of Sitemap Fetch and uses the message Sitemap no encontrado o acceso bloqueadp.

Step 8: Test and Activate Your Workflow

Run a controlled test to validate the end-to-end chat response and sitemap summarization flow.

1. Click Execute Workflow and send a chat message through Web Chat Trigger with a URL like https://example.com.
2. Confirm a valid URL routes through URL Valid Check to Scan Options, Robots.txt Request, and Sitemap Fetch.
3. Verify that Page Summary Model runs and Append Sheet Row adds a row with Lang, Page URL, and Summary Content.
4. Check that Agent Response returns a chat reply when Schema Check is true and that Invalid URL Reply triggers for invalid URLs.
5. Once validated, switch the workflow to Active for production use.