Step 2: Fetch and Extract Homepage Content

Pull the lead’s website homepage and extract structured HTML content for AI summarization.

1. Open Fetch Homepage Content and set URL to the expression:
   ={{\n(() => {\n // 1. Trim whitespace (Crucial first step)\n let url = $json.Website.trim();\n\n // 2. Add protocol if missing (Default to https)\n if (!url.match(/^https?:\\/\\//)) {\n url = 'https://' + url;\n }\n\n try {\n // 3. Parse the URL object\n const urlObj = new URL(url);\n\n // 4. Add 'www.' ONLY if:\n // a) It is missing\n // b) It looks like a root domain (has exactly 2 parts, e.g., 'harvex-ai.com')\n // This prevents breaking subdomains like 'docs.google.com'\n if (!urlObj.hostname.startsWith('www.') && urlObj.hostname.split('.').length === 2) {\n urlObj.hostname = 'www.' + urlObj.hostname;\n }\n\n // 5. Return the clean URL\n return urlObj.toString();\n\n } catch (error) {\n // If the input is total garbage, return original to let the node error out naturally\n return url;\n }\n})()\n}}
2. Ensure Fetch Homepage Content keeps Response Format set to text.
3. In Extract Homepage HTML, set Operation to extractHtmlContent and map:
4. Use extraction values: h1 = h1, h2 = h2 (return array), meta_description = meta[name="description"] (attribute content), nav_links = nav a, header a (attribute href, return array).

Step 3: Configure AI Summarization and Status Routing

Generate an initial summary and branch based on whether more data is needed.

1. Open AI Summary Draft and select model chatgpt-4o-latest.
2. Credential Required: Connect your openAiApi credentials in AI Summary Draft.
3. Confirm the user prompt includes the expression: =H1: {{ $json.h1 }} H2: {{ $json.h2 }} Description: {{ $json.meta_description }} Links: {{ $json.nav_links }}.
4. In Branch on AI Status, keep the rules that check ={{ $json.output[0].content[0].text }} for "status": "success" and "status": "need_more_info", with Fallback Output set to extra.

Step 4: Build Fallback Scrape and Deep Summary

When the homepage is unclear, fetch an “About” page and produce a deeper AI summary.

1. In Build Fallback Link, set Fallback URL to:
   ={{\n(() => {\n const extractedPath = $json.output[0].content[0].text.match(/\"fallback_url\":\\s*\"([^\"]+)\"/)[1];\n\n\n const rawBase = $('Form Intake Trigger').item.json.Website;\n\n\n const cleanBase = (rawBase.startsWith('http') ? rawBase : 'https://' + rawBase).replace(/\\/$/, \"\");\n\n // 4. Combine them\n return extractedPath.startsWith('http') ? extractedPath : cleanBase + extractedPath;\n})()\n}}
2. In Fetch Fallback Page, set URL to ={{ $json["Fallback URL"] }} and keep Response Format as text.
3. In Extract Fallback HTML, set Operation to extractHtmlContent and extract about_text from main.
4. Open AI Deep Summary, select model chatgpt-4o-latest, and confirm the user prompt includes =Page Text: {{ $json.about_text }}.
5. Credential Required: Connect your openAiApi credentials in AI Deep Summary.
6. In Capture Script Text, map final_script to ={{ $json.output[0].content[0].text }}.

Step 5: Parse Summary and Aggregate Lead Data

Normalize AI outputs into a unified payload before creating CRM records.

1. In Parse Summary JSON, set final_script to:
   ={{ $json.output[0].content[0].text.match(/\"summary\":\\s*\"([^\"]+)\"/)[1] }}
2. In Combine Lead Data, set Aggregate to aggregateAllItemData to merge the summary with form data.

Step 6: Configure CRM Creation and Slack Notifications

Save enriched lead records and alert the team.

1. Open Create CRM Entry and select your Airtable Base and Table (e.g., CRM Base and Lead Table).
2. Credential Required: Connect your airtableTokenApi credentials in Create CRM Entry.
3. Map columns exactly: Date = ={{ $('Form Intake Trigger').item.json.submittedAt }}, Name = ={{ $('Form Intake Trigger').item.json.Name }}, Email = ={{ $('Form Intake Trigger').item.json.Email }}, Website = ={{ $('Form Intake Trigger').item.json.Website }}, AI Insight = ={{ $json.data[0].final_script }}.
4. Open Post Slack Alert, set Authentication to oAuth2, select your channel, and keep Text set to:
   =New lead acquired. \nName: {{ $json.fields.Name }}\nEmail: {{ $json.fields.Email }}\nWebsite: {{ $json.fields.Website }}\nAI Insight: {{ $json.fields['AI Insight'] }}
5. Credential Required: Connect your slackOAuth2Api credentials in Post Slack Alert.

Step 7: Add Scrape Failure Logging

Ensure failed scrapes are captured for manual review.

1. Open Log Scrape Failure and select the same Airtable Base and Table as the CRM node.
2. Credential Required: Connect your airtableTokenApi credentials in Log Scrape Failure.
3. Map fields: Date = ={{ $('Form Intake Trigger').item.json.submittedAt }}, Name = ={{ $('Form Intake Trigger').item.json.Name }}, Email = ={{ $('Form Intake Trigger').item.json.Email }}, Website = ={{ $('Form Intake Trigger').item.json.Website }}, AI Insight = Scrape Failed - Check URL manually.
4. Confirm Fetch Homepage Content and Fetch Fallback Page use Continue On Error to route failures into Log Scrape Failure.

Step 8: Test and Activate Your Workflow

Validate the flow from form submission to CRM and Slack.

1. Click Execute Workflow and submit a test entry via Form Intake Trigger.
2. Confirm a successful run creates a record in Airtable via Create CRM Entry with a populated AI Insight.
3. Verify Post Slack Alert posts the message to the selected channel.
4. If the homepage is unclear, ensure the fallback path runs through Build Fallback Link, Fetch Fallback Page, and AI Deep Summary.
5. When everything works, toggle the workflow to Active for production use.