3.4: Email Integration
Picture this: It's Monday morning. You open your support inbox and see 47 new emails that came in over the weekend. Each one needs a thoughtful, personalized response. That's hours of work...
But what if your AI agent could handle it?
Here's what we're building: An email automation system that:
- Monitors your support inbox 24/7
- Reads incoming customer emails automatically
- Drafts professional, helpful responses using AI
- Waits for your approval before sending anything
- Sends the approved response back to the customer
The best part? You review, AI does the work. You stay in control, but save hours every day.
Let's build this together!
3.4.1: What You'll Build
Here's what your automated email support looks like in action:
Support Inbox - Automated Processing
3 New"Hi! We're interested in your Enterprise plan for our team of 50 people. Could you provide pricing details and what's included?"
"I tried to reset my password but the email never arrives. Can you help?"
"Hi John, I'd be happy to help! I've checked your account and manually sent a new password reset link to john@company.com. Please check your spam folder if you don't see it in a few minutes..."
Response sent automatically after approval
The workflow:
- Customer sends email → support@yourcompany.com
- Email appears in your taible → As a new row with all details
- AI drafts response → Searches your knowledge base, writes helpful reply
- You review → Check the AI draft, make sure it's perfect
- You approve → Click one checkbox
- Email sends automatically → Customer gets instant, professional help
💡 Why This Is Powerful
Traditional way: You write every email manually
- 47 emails × 5 minutes each = 4 hours of repetitive work
With AI automation:
- AI writes 47 drafts × 10 seconds each = 8 minutes
- You review 47 drafts × 30 seconds each = 24 minutes
- Total: 32 minutes vs 4 hours
You save 3.5 hours, and your responses are faster and more consistent!
3.4.2: Why Email Is Different
Before we build, let's understand why email automation needs special care.
💡 💡 Email vs Chat: The Key Difference
Chat widgets (from section 3.2):
- Instant, informal, real-time
- Mistakes can be quickly corrected
- Lower stakes conversations
Email:
- Permanent record
- Professional communication
- Represents your company brand
- Can't easily "take back" a sent email
That's why we add human approval for email!
Email is more formal and permanent than chat, so we want you to review every response before it goes out. This gives you:
- ✅ Quality control
- ✅ Brand consistency
- ✅ Peace of mind
- ✅ Ability to catch edge cases
3.4.3: Step 1 - Set Up Incoming Email
First, we need to tell your taible to watch your email inbox.
What we're doing: Creating an IMAP trigger that monitors your email account and creates a new row whenever an email arrives.
Add Email Trigger (IMAP)
Step 1 of 3This trigger monitors your email inbox and creates a new row whenever an email arrives. Perfect for automating shared inbox processing, invoice handling, and support requests.
The email account to monitor. Configure accounts in Settings → Accounts.
The mailbox folder to monitor (usually INBOX)
How often to check for new emails
Optional Filters
Leave empty to process all emails, or add filters to only process specific emails
Let's do it:
- Click "Add Trigger" button in your taible toolbar
- Find "IMAP Email" trigger in the list
- Look in the "Communication" category
- Or search for "IMAP" or "email"
- Select your email provider
- Gmail (most common)
- Outlook / Office 365
- Or custom IMAP server
- Click "Connect Email Account"
- A popup appears asking for permission
- Authorize Taibles to read your emails
- Select the account (e.g., support@yourcompany.com)
- Choose which folder to monitor
- Usually "INBOX"
- Or a specific folder like "Support" or "Sales"
- Click "Save Trigger"
💡 🔒 Security Note
What access does this need?
- Read emails from your inbox
- Mark emails as read (optional)
What it does NOT access:
- Send emails (that's a separate column we'll add later)
- Delete emails
- Access other folders (unless you specify)
You're always in control of permissions!
What happens now: When an email arrives at support@yourcompany.com, your trigger fires and creates a new row in your taible. The email details (sender, subject, body) are automatically captured.
3.4.4: Step 2 - Create the Email Columns
Now we need columns to store the email data and process it.
Setting Up Email Columns
Customer Email
Subject
Message Body
AI Response
Approved to Send?
Send Email Response
Let's create all six columns:
Column 1: Customer Email
- Click "+ Add Column"
- Choose "Text" type
- Name it:
customer_email - Label: "Customer Email"
- Click "Save"
This will automatically be filled by the email trigger with the sender's address.
Column 2: Subject
- Click "+ Add Column"
- Choose "Text" type
- Name it:
subject - Label: "Subject"
- Click "Save"
This stores the email subject line.
Column 3: Message Body
- Click "+ Add Column"
- Choose "Text" type (multi-line)
- Name it:
message_body - Label: "Customer Message"
- Click "Save"
This is the full email content from your customer.
Column 4: AI Response
Now for the AI magic!
- Click "+ Add Column"
- Choose "AI Assistant" type
- Found in "Talk to Customers" category
- Name it:
ai_response - Label: "AI Draft Response"
- Configure dependencies:
- Add dependency: "Message Body"
- This means AI waits for the message before responding
- Write instructions:
You are a helpful customer support assistant for [Your Company]. Read the customer's email and draft a professional, friendly response. Search our knowledge base for accurate information. Guidelines: - Be warm and professional - Answer their question completely - Offer to help with anything else - Keep it concise but thorough - Enable "Search Knowledge Base" (if you have one set up)
- Click "Save"
💡 Pro Tip: Customize Instructions
Customize the AI instructions for your business:
- Mention your company name
- Add your support policies
- Include common troubleshooting steps
- Set the tone (formal vs casual)
The better your instructions, the better the AI responses!
Column 5: Approved to Send?
This is your approval gate!
- Click "+ Add Column"
- Choose "Boolean" type (checkbox)
- Name it:
approved - Label: "Approved to Send?"
- Set default value: Unchecked (false)
- Enable manual entry: Yes
- Click "Save"
This checkbox lets you approve each email before it goes out.
Column 6: Send Email Response
Finally, the column that actually sends the email!
- Click "+ Add Column"
- Choose "Send Email (SMTP)" type
- Found in "Talk to Customers" category
- Name it:
send_response - Label: "Send Email"
- Configure email fields:
Configure Send Email Column
SMTPLet's set up each field:
To: (Who receives the email)
- Click in the "To" field
- Click "Customer Email" button to insert that column
- This ensures reply goes back to the original sender
Subject: (Email subject line)
- Type:
Re: - Click "Subject" button to insert the original subject
- Result: "Re: [Original Subject]"
Body: (The email content)
- Click "AI Response" button
- This inserts your AI-generated response
When to send: 6. Add condition: "Only run if..."
- Select column: "Approved to Send?"
- Operator: "equals"
- Value: "true" (checked)
- Click "Save"
💡 ⚠️ Critical: Add the Approval Condition!
This step is ESSENTIAL:
Without the approval condition, emails would send automatically without review. That's dangerous!
Always add the condition: "Only run if Approved to Send? equals true"
This ensures you review every email before it goes out.
3.4.5: Step 3 - Test the Workflow
Time to see it in action!
Let's test:
Send a test email to your monitored address
- From your personal email
- To: support@yourcompany.com
- Subject: "Test: Need help with billing"
- Body: "Hi, I need help understanding my invoice. Can you explain the charges?"
Watch your taible (refresh if needed)
- A new row appears!
- Customer Email, Subject, and Message Body fill in automatically
Wait for AI response (usually 5-10 seconds)
- The "AI Draft Response" column populates
- Read the AI-generated response
Review the AI draft
- Is it accurate?
- Is the tone appropriate?
- Does it answer the question completely?
Approve by checking the box
- Click the checkbox in "Approved to Send?" column
- The checkbox turns green (checked)
Watch the email send (happens in 1-2 seconds)
- The "Send Email" column shows "Sent ✓"
- Check your personal email inbox
- You should receive the AI response!
💡 🎉 You Did It!
Congratulations! You just automated email support with AI assistance!
Your system now:
- ✅ Monitors your inbox 24/7
- ✅ Creates rows for every email
- ✅ Drafts intelligent responses
- ✅ Waits for your approval
- ✅ Sends emails automatically when approved
You're in control, but AI does the heavy lifting!
3.4.6: The Approval Workflow in Detail
Let's understand exactly how the approval process works:
Email Approval Workflow
Step 1: Email Arrives
CompleteStep 2: AI Drafts Response
CompleteStep 3: You Review & Approve
Waiting for You✓ Information is accurate
✓ Tone is professional
Step 4: Email Sends Automatically
PendingThe sequence:
Step 1: Email arrives → Row created → Email details captured Step 2: AI processes → Searches knowledge → Generates response Step 3: You review → Read the draft → Decide if it's good Step 4: You approve → Check the box → Triggers the send Step 5: Email sends → Customer receives → Support complete
💡 💡 Why This Works So Well
Human + AI collaboration:
AI strengths:
- Speed (instant drafts)
- Consistency (same quality every time)
- Knowledge (searches all your docs)
- Availability (24/7, never tired)
Your strengths:
- Judgment (edge cases and exceptions)
- Empathy (emotional intelligence)
- Brand voice (maintaining your style)
- Final approval (quality control)
Together, you're unstoppable!
3.4.7: Customizing for Your Needs
Now that the basics work, let's customize!
Add Email Filters
Want to only process certain emails?
In the IMAP trigger settings:
- Subject filter: Only emails with "Support" in subject
- From filter: Only emails from @companydomain.com
- Skip seen emails: Don't process emails you've already read manually
Customize AI Responses
Make the AI sound like YOU:
Edit the "AI Response" column instructions:
- Add your greeting style ("Hey there!" vs "Dear customer")
- Include your sign-off ("Cheers!" vs "Best regards")
- Add policies (refund policy, response times, etc.)
- Mention your name or team name
Add Priority Handling
Want urgent emails handled faster?
Add a "Priority" column:
- Type: AI Classification
- Classifies emails as: "Urgent", "Normal", "Low"
- Based on keywords, tone, customer status
Then you can:
- Filter view to see "Urgent" emails first
- Send notifications for urgent emails
- Apply different response templates by priority
Multiple Email Addresses
Monitoring multiple inboxes?
Create separate taibles for each:
- "Sales Support" taible → sales@company.com
- "Technical Support" taible → support@company.com
- "Billing" taible → billing@company.com
Or use one taible with a "Department" column:
- Set up multiple IMAP triggers
- Each trigger fills the "Department" column differently
- Use filters to separate them
3.4.8: Best Practices
🎯 Always review before approving:
- Read every AI response carefully
- Check facts and information
- Verify tone matches your brand
- Make sure it fully addresses the question
🚀 Speed up reviews:
- Use keyboard shortcuts
- Set up filters for different email types
- Batch-approve similar emails
- Create templates for common scenarios
📊 Monitor quality:
- Periodically review sent emails
- Collect customer feedback
- Refine AI instructions based on performance
- Update knowledge base when you find gaps
🔒 Maintain security:
- Never approve emails with sensitive data if uncertain
- Review permissions regularly
- Use separate email accounts for different purposes
- Keep your knowledge base up-to-date
3.4.9: Troubleshooting
"Emails aren't creating rows"
- Check your IMAP trigger is enabled
- Verify email account is still connected
- Check folder name is correct ("INBOX" vs "Inbox")
- Look for authorization errors
"AI responses are generic"
- Add more detailed instructions
- Set up a knowledge base with your FAQs
- Provide examples of good responses in instructions
- Reference specific policies or documentation
"Approved emails aren't sending"
- Verify "Send Email" column has the approval condition
- Check your SMTP account is connected
- Make sure the "To" field references the right column
- Look for error messages in the "Send Email" column
"Too many test emails in my inbox"
- Use a separate test email account
- Create a filter to auto-archive test emails
- Or use a separate taible for testing
What You've Learned
Amazing work! You've just built a complete email automation system.
You now know how to:
- ✅ Set up an IMAP trigger to monitor email
- ✅ Create columns for email data and AI responses
- ✅ Configure AI to draft professional responses
- ✅ Add human approval gates
- ✅ Automatically send approved emails
- ✅ Customize the workflow for your needs
The power of this:
- Your support inbox runs 24/7
- AI drafts responses instantly
- You review and approve (stay in control)
- Customers get fast, consistent help
- You save hours every single day
Next up: In section 3.5, we'll connect your agent to Slack so your team can get instant help from AI in their workspace!
Previous: 3.3 WhatsApp IntegrationNext: 3.5 Slack Integration