Day 5: AI responder & lead qualifier – Full automation

Day1

prompts

Day2

Zapier logic

Day3

Make multi-step

Day4

OpenAI API

Day5

AI lead responder

Capstone: All skills combined into a real business system – intelligent email processing with zero manual work.

🏆 The complete AI lead machine

📌 System overview

You'll build an automation that:

Watches a Gmail inbox (e.g., [email protected]).
Qualifies the lead using AI: budget, timeline, interest (Day1 prompt).
Logs the lead in Google Sheets (Day2).
Routes hot leads to Slack (Day3).
Generates a personalised reply via OpenAI API (Day4).
Sends the reply automatically (or stores draft).

Real-world analogy: It's like hiring a 24/7 sales assistant who reads every email, identifies promising leads, writes a custom response, and alerts your team for the hot ones.

📊 How the pieces fit together

1. Trigger

Gmail (new email)

2. Qualify

OpenAI API call (Day4) with BANT prompt

3. Router

Hot / Warm / Cold (Day3)

4. Log

Google Sheets (Day2)

5. Generate reply

Second OpenAI call (personalised)

6. Send / Draft

Gmail reply

⚙️ Build the lead qualifier & responder (Make.com)

We'll use Make (from Day3) and OpenAI API (Day4) for full control.

Trigger: Gmail – Watch new emails

Select inbox/label (e.g., "leads"). Get sample email.

Reuse Day2: same trigger concept.

Qualify lead – OpenAI API call (BANT)

Add HTTP – Make a request (like Day4). Use this prompt (refined from Day1):

{
  "model": "gpt-3.5-turbo",
  "messages": [
    {"role": "system", "content": "You are a BANT qualifier. Extract: Budget (high/medium/low), Authority (yes/no), Need (strong/weak), Timeline (soon/medium/later). Return JSON."},
    {"role": "user", "content": "Email: {{subject}} {{body plain}}"}
  ],
  "max_tokens": 150
}
                        

Day1 mastery: This prompt is precise and asks for JSON – exactly what we learned.

Parse JSON & set lead score

Add JSON – Parse module. Map fields: budget, authority, need, timeline.

Then add a variable module to calculate a score (e.g., high=3, medium=2, low=1). Score >=8 = hot.

Formula example: {{if(parse.budget="high";3;if(parse.budget="medium";2;1)) + ...}}

Router: hot / warm / cold

Add a router with three paths based on score:

🔥 Hot (score ≥8) → Slack + high-priority sheet + send reply
👍 Warm (4-7) → sheet only + send reply
❄️ Cold (≤3) → sheet only (maybe no reply)

Log to Google Sheets (all leads)

On every route, add Google Sheets – Add row. Map: timestamp, from, subject, BANT fields, score, reply status.

Day2 in action: you're now logging enriched data.

Generate personalised reply (second AI call)

For hot/warm paths, add another HTTP call to OpenAI with a different prompt:

{
  "messages": [
    {"role": "system", "content": "Write a friendly, helpful reply to this lead. Thank them, address their need briefly, and suggest a next step (call/demo). Sign as 'Alex from [Company]'."},
    {"role": "user", "content": "Original email: {{subject}} {{body plain}}"}
  ]
}
                        

Send reply or create draft

Add Gmail – Send email module. Use:

To: {{trigger.from}}
Subject: Re: {{trigger.subject}}
Body: {{ai_reply.content}}

Safety: For testing, use "Create draft" instead of send.

Slack alert for hot leads

On hot path, add Slack – Send message with lead details and AI summary.

🚨 *Hot lead alert!* From: {{from}} Score: 9/10 Need: {{parse.need}} Timeline: {{parse.timeline}}

🧩 Most complex part – made simple

"Two AI calls in one scenario? How do I handle different prompts and responses?"

Treat each AI call as a separate "expert". First expert = qualifier (reads email, outputs JSON). Second expert = writer (reads email + score, outputs reply). They don't mix; you just pass the email to both.

Visual: Email → [Qualifier AI] → JSON → [Router] → [Writer AI] → reply. Data flows step by step.

⚠️ Critical safety tips (read carefully)

Always test with drafts first – use Gmail's "Create draft" before "Send".
Add a human review step for hot leads: send to Slack and let a human approve the reply (use Make's "waiting" module).
Set a sender name that makes sense – avoid robotic signatures.
Monitor costs: two AI calls per email. Use gpt-3.5-turbo and low tokens.

Practice to master

Basic qualifier

Build only the qualification part: email → AI → sheet (no reply). Use a simple "hot/warm/cold" prompt.

Add reply (draft)

Extend with second AI call and Gmail draft. Test with your own email.

Slack + human approval

Add a "human in the loop": hot lead posts to Slack with Approve/Reject buttons (Make's webhook). Only send reply if approved.

📚 Resources for deeper lead qualification

Day 5: You built a complete AI lead qualification & response system

✔ Integrated Gmail, OpenAI API (twice!), Sheets, Slack, Router
✔ Reused prompts from Day1, Zapier logic from Day2, Make skills from Day3, API calls from Day4
✔ Understood safety & human oversight
✔ Created a real business asset – 24/7 lead machine

You've graduated from beginner to automation architect.