Day 3: Make.com multi‑step – Connect Gmail, Sheets, Slack & AI

Day 1

prompt engineering (clarity, structure)

Day 2

Zapier: Gmail→Sheets (trigger → action)

Day 3

Make.com multi‑step: routing, AI, Slack

Synthesis: Day1 precision + Day2 logic = Day3 complex, conditional, multi‑app scenarios.

🔀 What is Make.com (formerly Integromat)?

📌 Visual automation platform

Make.com lets you build scenarios that connect apps via a visual editor. Unlike Zapier's linear steps, Make offers routers, iterators, data stores, and deep conditional logic. Each “scenario” is a flowchart of modules – you see data flowing in real time.

Analogy: If Zapier is like a simple conveyor belt (Day2), Make.com is a fully programmable assembly line with switches, forks, and quality control. You can add AI (Day1) to decide where each email should go.

Trigger

Watch emails (Gmail)

Router

split by subject / content

AI module

OpenAI (prompt from Day1)

Sheets

log data

Slack

send notification

Filter

only important emails

🏢 Scenario: Support ticket triage with AI

1. Email arrives

[email protected] (Gmail trigger)

2. AI categorises

OpenAI module (Day1 prompt) extracts urgency, department

3. Router

high urgency → Slack + phone notify; low urgency → Sheets only

This goes far beyond Day2's simple copy – you're adding intelligence and branching.

⚙️ Build: AI‑powered email router (Gmail → OpenAI → Sheets/Slack)

Create a new scenario & add Gmail trigger

In Make.com dashboard: Create a new scenario. Click the big + and search for Gmail. Choose "Watch emails" (or "Watch new emails"). Connect your Gmail account.

Day2 connection: same trigger concept – but now you can set a custom search query using Day1 prompt thinking: subject:"support" OR label:inbox

Add OpenAI module – analyse sentiment & urgency

Add a new module: choose OpenAI – Create Chat Completion. Use a system prompt crafted with Day1 skills.

🔹 System prompt: You are a support classifier. Return JSON with "urgent"(true/false) and "department" (billing/tech/sales).  
🔹 User prompt: Subject: {{subject}} Body: {{body plain}}  
                        

This is Day1 in action – the prompt engineering determines the quality of automation.

Parse JSON & Router

Add a JSON module to parse the OpenAI response. Then add a Router. Create two routes:

Route A (urgent) → Slack message + high‑priority row in Sheets
Route B (non‑urgent) → only log to Sheets (like Day2)

Router filters: {{json.urgent}} = true (route A) else route B.

Google Sheets – log all emails

On both routes, add a Google Sheets – Add a row module. Map: Timestamp, From, Subject, AI category, Urgency flag.

Day2 reminder: field mapping is identical – but now you also map AI‑generated fields.

Slack notification (urgent only)

On the urgent route, add a Slack – Send a Message module. Compose a message with the email details and AI summary.

🚨 *Urgent support request*  
From: {{from}}  
Subject: {{subject}}  
Department: {{json.department}}  
                        

Test, schedule & publish

Run the scenario once with a sample email. Check the data flow visually. Set schedule (e.g., every 5 minutes) and turn scenario ON.

✅ You've now built a multi‑step, AI‑augmented workflow that surpasses simple Zapier transfers.

Scenario visual flow

[Gmail] → [OpenAI] → [JSON parse] → [Router] 
                                      ↙️ (urgent) ↘️
                              [Sheets+Slack]  [Sheets only]

🧩 Make.com pro tips & troubleshooting

Not parsing JSON from OpenAI → can't access fields
Router without a fallback route → emails disappear
Forgetting to add error handlers (scenario breaks)

Always use JSON module or set "parse response" in OpenAI
Add a catch‑all route (default) to the router
Attach error handler modules to critical steps

Day1 again: The OpenAI prompt must ask for valid JSON – otherwise parsing fails. Precise instructions avoid broken automations.

Practice scenarios

Simple router

Create a scenario that watches Gmail, uses a filter (not AI) to split by subject "invoice" → Sheets A, other → Sheets B.

AI sentiment

Add OpenAI to classify emails as positive/negative and send only negative to Slack (Day1 prompt: "sentiment analysis").

Data store

Use Make's data store to count emails per sender – if >5 per day, send alert (advanced).

📚 Make.com resources & deep dives

Day 3: you mastered multi‑step Make.com scenarios

✔ Combined Gmail, OpenAI (Day1), Sheets (Day2), and Slack
✔ Used routers for conditional branches
✔ Understood JSON parsing & error handling
✔ Connected all three days into a production‑grade automation