The Live Build: The Researching Lead Bot

Today, we build a real-world system from scratch. By the end of this lesson, you'll have a working automation that researches leads for you automatically.

The Project: Automated Lead Research Pipeline

Goal: When someone adds a company name and website to a Google Sheet, automatically: 1) Research that company, 2) Draft a personalized email icebreaker, 3) Send you a Slack notification with everything.

Step 1: Set Up Your Google Sheet

Create a sheet with columns:

• Column A: Timestamp
• Column B: Company Name
• Column C: Website URL
• Column D: Contact Name (optional)
• Column E: Research Summary (AI fills this)
• Column F: Icebreaker Email Draft (AI fills this)
• Column G: Status (Pending / Approved / Sent)

Step 2: Build the Make.com Automation

1. Trigger: New row added to Google Sheet (watch for new rows)
2. Step: Get Company Name and Website URL from the row
3. Step: Perplexity/OpenAI searches for company news, funding, and "About Us" info
   • Prompt: "Research [Company Name] at [Website]. Find: 1) What they do, 2) Latest news or updates, 3) Key products or services, 4) Any recent achievements or announcements."
4. Step: OpenAI generates the email icebreaker
   • Prompt: "Based on this research: [Paste research]. Write a 100-word personalized LinkedIn connection request or email icebreaker. Reference something specific from the research. Sound human, not AI-generated. Keep under 120 words."
5. Step: Send Slack message to your channel with: company name, research summary, email draft, link to approve
6. Step: Update Google Sheet row with research summary and draft email

Step 3: Setting the Temperature

This is critical — different tasks need different AI creativity levels:

• Temperature 0.1-0.2 (Very low): Research and factual extraction — you want accuracy, not creativity
  • "What is their revenue?" / "What did they announce last week?"
• Temperature 0.5-0.7 (Medium): Creative tasks like email drafts — you want it to sound human
  • "Write a connection request email" / "Draft a follow-up message"
• Temperature 0.8-1.0 (High): Brainstorming, creative writing, when you want unexpected ideas

Step 4: Error Troubleshooting

• If the automation stops: Check whether your API credits have run out or expired (free credits usually do). Check Make.com operation log.
• If research is wrong: Your prompt is too vague. Add more constraints. Example: "Only use information from the company's official website or recent news articles (2023-2024). Do not speculate."
• If email sounds AI: Increase temperature to 0.7 and add: "Write like a human who did real research. No corporate speak."
• If Slack message is blank: Check that the Google Sheet columns are correctly mapped in Make.com

Step 5: The Approval Workflow

Don't auto-send emails. Instead:

1. Slack message includes "Approve: Reply YES to send, NO to discard"
2. You reply YES → Make.com sends the email and updates sheet to "Sent"
3. You reply NO → Make.com updates sheet to "Discarded" and does nothing

Step 6: Cost Management

• Track how many AI calls you're making
• Add a filter: only research if Company Name is not empty
• Add a daily limit: max 50 research runs per day
• Each research call costs a small amount that varies by model and by how much text you send — check your provider's current rate

⚠️ Common Mistakes

• Hardcoding values: Putting a specific company name in the prompt instead of using the "Variable" from the Google Sheet. Always use variables.
• Ignoring API costs: Running 1,000 research tasks without checking what one costs first. Multiply before you run, not after. It costs ₹10. Budget for this.
• Skipping the review step: Auto-sending AI-generated emails without human review can backfire badly. Always review first.
• Not documenting: Take screenshots of your automation. When it breaks (and it will), you'll thank yourself.

Advanced Build — The Full Pipeline:

1. Company added to Sheet → AI researches
2. Research shows "recent funding" → Flag for special outreach
3. Research shows "no recent news" → Use different email template
4. All results → Slack to you for review
5. You approve → Email sent → Sheet updated to "Sent"

What's Next: You are now an AI Automator. But what if the "automation" needs to live on WhatsApp?