Agentic AI Automation: Build AI Workflows Without Coding

Watch the video below from 11:15 to 15:36.

Focus on:

• The canvas (where you build workflows)

• The node panel (on the left)

• The settings panel (on the right)

• How to zoom and navigate

Watch the video from 15:36 to 28:20.

You'll learn:

• What a trigger is (hint: it starts your workflow)

• What action nodes do

• How to connect nodes together

STEP 1: Import the Template

• Click the link below- Click "Use this workflow"

• It will open in your n8n

STEP 2: Analyze WITHOUT Running Look at each node and write down:

• What is the trigger? When does it run?

• What does the HTTP Request node do?

• What does the Set node do?

• What does the Gmail node do?

• What will the final result be?

STEP 3: Predict the Outcome

• Write a paragraph describing what this workflow does in plain English.

STEP 4: Run and Verify

• Now execute the workflow and see if you were right!

We're starting with webhook triggers - one of the most powerful trigger types.

Watch the video from 1:43:19 to 1:46:22

PRACTICAL EXERCISE Assignment: Create 3 JSON-Mode Prompts

1. Product Review Analyzer

{

}

Input:

Customer review text

Output JSON:

{

  "rating_guess": 1-5,

  "sentiment": "positive/negative/neutral",

  "key_points": ["point1", "point2"],

  "would_recommend": true/false

}

2. Meeting Action Item Extractor

{

}

Input:

Meeting notes

Output JSON:

{

  "action_items": [

    {

      "task": "description",

      "assigned_to": "person",

      "due_date": "date if mentioned"

    }

  ]

}

3. Lead Qualifier

Input:

Contact form submission

Output JSON:

{

}

{

  "lead_quality": "hot/warm/cold",

  "budget_indicated": "yes/no/unclear",

  "timeline": "immediate/this_month/this_quarter/unclear",

  "recommended_action": "call_immediately/send_info/add_to_nurture"

}

Submit:

• Screenshots of all 3 working nodes with JSON output

• Your prompt text for each

PRACTICAL EXERCISE Assignment: Build an AI-Powered Workflow

Create a workflow that:

1. Webhook Trigger

Receives product reviews

{

  "product": "product name",

  "review": "review text",

  "rating": 1-5

}

2. OpenAI Node

Analyzes the review

Returns JSON with sentiment analysis and key themes

3. IF Node

Routes based on sentiment

• Negative → Send alert email

• Positive → Log to "Good Reviews" sheet

• Neutral → Log to "All Reviews" sheet

4. Multiple Action Nodes

Different actions for each path

Submit:

• Screenshot of the complete workflow

• Screenshot of 3 test executions (negative, positive, neutral reviews)

• Execution logs showing AI output and routing decisions

PRACTICAL EXERCISE Assignment: Build and Customize

1. Build the Email Summarizer workflow exactly as shown

2. Test it manually

(don't wait for scheduled time)

3. Make at least 2 customizations:

• Change the schedule time

• Modify the AI prompt

• Change query filters

• Add a new output destination

4. Let it run for 2 days

5. Submit:

• Screenshot of complete workflow

• Screenshot of received digest email

• Description of your customizations

• Feedback: How did it help you?

PRACTICAL EXERCISE Assignment: Build Message Organizer

1. Set up a Telegram bot

(or use WhatsApp if you have access)

2. Build the complete Message Organizer workflow

3. Create the Google Sheet with proper headers

4. Test with at least 5 different messages:

• One urgent work message

• One personal message

• One info/news message

• One spam-like message

• One that requires action

5. Submit:

• Screenshot of workflow

• Screenshot of Google Sheet with logged messages

• Screenshot of an urgent alert being sent

• One customization you added

PRACTICAL EXERCISE Assignment: Generate and Publish

1. Build the LinkedIn Post Generator

2. Set up Notion database

(or use Google Docs alternative)

3. Generate 3 posts on different topics:

• One professional tone

• One casual tone

• One inspirational tone

4. Edit and publish at least ONE to LinkedIn

5. Submit:

• Screenshots of workflow

• Screenshots of generated posts in Notion

• Link to published LinkedIn post (if comfortable)

• Feedback on AI quality

PRACTICAL EXERCISE Assignment: Build 3 Conditional Workflows

1. IF Node Practice:

• Manual trigger with Set node:

{

  "score": 85

}

• IF: score > 80

TRUE path:

Set node "Pass"

FALSE path:

Set node "Fail"

2. Switch Node Practice:

• Manual trigger with Set node:

{

  "day": "Monday"

}

• Switch based on day

Route 0:

Monday → "Start of week"

Route 1:

Friday → "End of week"

Route 2:

Saturday/Sunday → "Weekend"

Fallback:

"Midweek"

3. Combined Conditions:

• Manual trigger:

{

  "temperature": 28,

  "raining": false

}

• IF: (temperature > 25 AND raining = false)

TRUE →

"Perfect beach day!"

FALSE →

"Stay inside"

Submit:

Screenshots of all 3 workflows executing.

PRACTICAL EXERCISE Assignment: Build AI Decision Workflow

Build a Content Moderation System:

Pattern: First AI Decision → Second AI Decision → Action

1. Webhook receives:

{

  "content": "user-submitted text"

}

2. OpenAI analyzes for:

• is_appropriate (true/false)

• category (spam/offensive/safe/promotional)

• confidence (0-100)

3. IF is_appropriate = true AND confidence > 80:

TRUE →

Approve (add to database)

FALSE →

Flag for human review

4. Test with 3 different content samples

Submit:

• Screenshot of workflow

• Your OpenAI prompt

• 3 test executions with different results

PRACTICAL EXERCISE Assignment: Build Multi-Path Router

Create a Customer Inquiry Router:

1. Webhook receives:

{

  "inquiry_type": "...",

  "message": "..."

}

2. OpenAI analyzes and returns:

3. Switch based on department

(3 routes)

4. Each route:

Sales →

• CRM

• Sales email

Support →

• Ticket system

• Support email

Billing →

• Accounting sheet

• Billing email

5. Test all 3 paths

Submit:

• Screenshots of all 3 successful route executions

PRACTICAL EXERCISE

Assignment: Identify Failure Points

Review one of YOUR existing workflows and:

1. List 5 potential failure points

2. For each, identify:

• What could go wrong?

• Why would it fail?

• What would happen if it fails?

3. Rate severity

• Critical

• Medium

• Low

4. Propose a fix for each

Submit:

• Workflow screenshot with failure points marked

• Analysis document

PRACTICAL EXERCISE

Assignment: Add Error Handling

Take any existing workflow and add:

1. Error Trigger node with email notification

2. Identify 2 critical nodes → Enable "Continue on Fail"

3. Add IF checks after those nodes

4. Create error vs success paths

5. Test by intentionally causing errors

Submit:

• Before/after screenshots

• Screenshot of error email received

• Test execution logs showing error handling working

PRACTICAL EXERCISE

Assignment: Build Comprehensive Validation Create a validation workflow for a contact form: 1. Webhook receives: name, email, phone, message, budget

2. Code node validates all fields:

• Name: 2-50 chars

• Email: valid format

• Phone: optional, but if present, valid format

• Message: 10-500 chars

• Budget: number, minimum $500

3. IF validation passes → Continue

4. IF validation fails → Send error with specific issues

5. Test with 3 scenarios (all valid, missing field, invalid format)

Submit:

• Validation code

• Screenshots of all 3 test scenarios

• Error messages for failed validations

PRACTICAL EXERCISE

Assignment: Multi-Layer Fallback

Build a notification system with 3 fallback layers:

1. Try Slack notification

2. If fails → Email

3. If fails → Google Sheet log

4. Include timestamp and failure reason at each level 5. Test by disconnecting credentials

Submit:

• Workflow screenshot

• Test showing all 3 levels being used

• Log showing which fallback was triggered

PRACTICAL EXERCISE

Assignment: Analyze AI Builder Capabilities

1. Go to n8n AI Workflow Builder

2. Try 3 different prompts:

Simple:

"Send me email every morning"

Medium:

"Form submission to Google Sheets with email notification"

Complex:

"Customer support ticket routing system"

3. For each, observe:

• What nodes did it choose?

• How are they connected?

• What's missing?

• What's configured well?

Submit:

• Screenshots of all 3 generated workflows

• Analysis of strengths/weaknesses for each

PRACTICAL EXERCISE

Assignment: Generate 3 Workflows

1. Simple:

"Daily weather report via email"

2. Medium:

"Contact form to CRM with validation"

3. Complex:

"Content approval system with AI review and multi-stage routing"

For each:

• Write your prompt

• Generate the workflow

• Screenshot the result

• List what's good and what's missing

Submit all documentation.

PRACTICAL EXERCISE

Assignment: Audit an AI Workflow

1. Generate a workflow with prompt: "Create a workflow that processes customer support

tickets from email, categorizes them with AI, and routes to appropriate team members"

2. Perform complete 7-point audit:

Document findings for each point

List what works

List what's missing

Rate production readiness (%)

3. Create fix plan:

Prioritize fixes (Critical/Important/Nice-to-have)

Estimate time for each fix

Submit:

Generated workflow screenshot

Complete audit report

Fix plan document

PRACTICAL EXERCISE

Assignment: Complete Fix Cycle

1. Use the workflow you audited in 8.3

2. Fix ALL critical issues

3. Fix at least 2 important issues

4. Add 1 nice-to-have improvement

5. Test all paths

6. Document before/after

Submit:

Before screenshot (AI-generated)

After screenshot (your fixes)

List of all fixes made

Test results showing it works