Automate Your Email Inbox with Make.com + AI

Email overload is real. The average professional receives 120+ emails per day and spends 2.5 hours managing their inbox. What if AI could handle 80% of that work for you?

What You Will Build:

• Auto-categorize emails (Urgent, Important, FYI, Newsletter)
• AI-powered priority scoring
• Draft responses for common inquiries
• Auto-archive newsletters and promotions
• Daily digest of important emails only

What You Will Need

Make.com account (free tier works - 1,000 operations/month)
OpenAI API key (GPT-3.5 costs ~$0.01 per email analyzed)
Gmail or Outlook account
30 minutes to set up

Expected cost: For 100 emails/day, you will spend about $3/month on AI and $0 on Make.com (free tier).

Step 1: Set Up Make.com

1.1 Create Your Make.com Account

Go to make.com and sign up for free
Choose the Free plan (1,000 operations/month is plenty to start)
Verify your email and complete onboarding

1.2 Connect Your Gmail Account

Click "Create a new scenario"
Search for "Gmail" in the apps list
Click "Watch emails" trigger
Click "Add" next to Connection
Authorize Make.com to access your Gmail (read-only is fine)

Security Note: Make.com only gets access to the permissions you grant. For this tutorial, we only need "Read" access to emails in specific folders.

Step 2: Configure the Email Trigger

In the Gmail module, configure these settings:

Folder: INBOX

Criteria: Unread messages

Maximum number of results: 10

Mark message(s) as read: Yes

This will check your inbox every 15 minutes (default) and process up to 10 new emails per run.

Step 3: Add OpenAI for Email Analysis

3.1 Get Your OpenAI API Key

Go to platform.openai.com/api-keys
Create a new secret key (copy it immediately - you won't see it again!)
Add $5-$10 in credits to your account

3.2 Add OpenAI Module in Make.com

Click the "+" button after your Gmail module
Search for "OpenAI"
Select "Create a Completion"
Add your API key when prompted

Step 4: The AI Categorization Prompt

This is the secret sauce. Here's the exact prompt to use in the OpenAI module:

You are an email categorization assistant. Analyze this email and return ONLY a JSON object.

Email Details:
From: {{1.from}}
Subject: {{1.subject}}
Body: {{1.textPlain}}

Respond with this exact JSON structure:
{
  "category": "urgent|important|fyi|newsletter|spam",
  "priority": 1-10,
  "action_needed": true/false,
  "suggested_response": "draft response or null",
  "reasoning": "brief explanation"
}

Rules:
- "urgent": needs response within 24h (meetings, deadlines, client issues)
- "important": needs attention but not urgent (project updates, reports)
- "fyi": informational only (updates, announcements)
- "newsletter": marketing emails, subscriptions
- "spam": unsolicited, irrelevant
- priority 1 = lowest, 10 = highest urgency
- suggested_response: only if it's a simple question that can be answered in 1-2 sentences

OpenAI Module Settings:

Model: gpt-3.5-turbo

Max Tokens: 500

Temperature: 0.3 (lower = more consistent)

Step 5: Parse the AI Response

Add a "JSON > Parse JSON" module after OpenAI:

Click "+" after OpenAI module
Search for "JSON"
Select "Parse JSON"
Map the OpenAI response: {{2.choices[].message.content}}

Step 6: Route Based on Category

Now add a Router to handle different email types:

Route 1: Urgent Emails

Condition: category = "urgent"
Action: Add Gmail label "🔴 URGENT"
Action: Star the email
Action: Send yourself a Slack/SMS notification (optional)

Route 2: Important Emails

Condition: category = "important"
Action: Add Gmail label "🟡 Important"
Keep in inbox for review

Route 3: FYI/Newsletters

Condition: category = "fyi" OR "newsletter"
Action: Add label "📰 FYI"
Action: Archive email (move out of inbox)

Route 4: Spam

Condition: category = "spam"
Action: Move to Trash

Step 7: Auto-Draft Responses (Optional)

For emails where AI suggested a response, you can auto-save drafts:

Add another Router after the JSON parse
Condition: suggested_response != null
Add "Gmail > Create a Draft" module
To: {{1.from}}
Subject: Re: {{1.subject}}
Content: {{3.suggested_response}}

Pro Tip: Drafts will appear in your Gmail Drafts folder. You can review and send them manually, or set up another automation to auto-send simple responses.

Step 8: Test Your Automation

Before going live:

Run Once: Click "Run once" in Make.com to test with real emails
Check Output: Review the AI categorizations - are they accurate?
Adjust Prompt: Refine the categorization rules if needed
Monitor Costs: Check OpenAI usage (should be ~$0.01-0.02 per email)

Step 9: Daily Digest (Bonus)

Want a daily summary of important emails? Add this final step:

Create a new scenario with a "Schedule" trigger (once per day at 6 PM)
Add "Gmail > Search emails" module
Search query: label:important OR label:urgent newer_than:1d
Add "OpenAI > Create Completion" to summarize all emails
Use prompt: "Summarize these emails into a brief daily digest with action items"
Send the summary to yourself via Gmail/Slack

Results You Can Expect

80%

of emails auto-categorized correctly

45 min/day

saved on email management

$3/month

total cost for 100 emails/day

Zero

missed urgent emails

Troubleshooting

AI is miscategorizing emails

Adjust your prompt to include more examples. Add a few sample emails with their correct categories in the prompt. Also try increasing temperature to 0.5 for more nuanced decisions.

Scenario runs but nothing happens

Check that you marked messages as "read" in the Gmail trigger. Otherwise, it will keep processing the same emails. Also verify your Router conditions are correctly formatted.

OpenAI costs are too high

Limit the email body to first 1000 characters in the prompt. Most categorization can be done from subject + first paragraph. Also, use gpt-3.5-turbo instead of gpt-4.

Make.com free tier limit reached

Each email processed = ~5 operations (trigger, AI call, parse, route, label). With 1,000 operations/month, you can handle ~200 emails. Either upgrade to Core ($9/month for 10K ops) or reduce checking frequency to once per hour instead of every 15 min.

Next Steps

Now that you have basic email automation working, here are some advanced features to add:

Sentiment Analysis: Flag angry/frustrated customer emails for immediate attention
Auto-Scheduling: If email contains meeting request, extract times and create calendar event
CRM Integration: Automatically log customer emails to HubSpot/Salesforce
Smart Replies: Train AI on your writing style for better draft responses
Attachment Processing: Extract data from PDFs/invoices and add to spreadsheet

Download the Make.com Template

Want to skip the manual setup? We've created a pre-built Make.com blueprint you can import.