🚀 Deployment and Maintenance - Crafter.ai

This comprehensive guide will walk you through the process of publishing your chatbot and managing it over time, including updates, maintenance, and monitoring.

🎯 Overview of the Deploy Step

The Deploy step is the final phase where:

• 🔧 You verify the prerequisites for publication
• 🚀 You publish the chatbot on production servers
• 🔗 You obtain the code for integration into the site
• 🔄 You manage updates and maintenance

✅ Prerequisites for Deployment

Mandatory Technical Checks

1. Training Completed

✅ Status: "Trained"
✅ Knowledge Base: Populated and processed
✅ Q&A Base: Minimum 10 active Q&As
✅ Survey: 100% completed

2. Active and Valid Plan

✅ Subscription: Active and not expired
✅ Message Limit: Not reached (< 100% usage)
✅ Payment Status: Updated
✅ Team License: Valid for deployment

3. Complete Configuration

✅ Bot Info: Name and description set
✅ Features: Functionality configured
✅ Appearance: Customization completed
✅ Settings: Behaviors configured

Possible Pre-Deploy States

🟢 Ready to Deploy

All prerequisites are met

• "Deploy" button active
• Process can start immediately
• Estimated time: 3-8 minutes

🟡 Needs Update (Outdated)

Existing chatbot but modified

• Changes to the knowledge base
• Updates to Q&As
• New chatroom configurations
• Action: Click "Update Bot"

🔴 Cannot Deploy

Prerequisites not met

Possible causes and solutions:

📚 Incomplete Survey

Error: "Incomplete survey"
Solution: 
1. Go to the Teaching Step
2. Complete all survey questions
3. Save and return to Deploy

🎓 Bot Not Trained

Error: "The chatbot is not trained"
Solution:
1. Go to the Teaching Step  
2. Click "Train Bot"
3. Wait for completion
4. Return to Deploy

💳 Plan Issues

Error: "Upgrade chatbot plan"
Solution:
1. Check active plan
2. Check message limits
3. Update payment if necessary
4. Contact team administrator

⚠️ Message Limit Reached

Error: "Message usage limit reached"
Solution:
1. Upgrade current plan
2. Swap to available license
3. Wait for monthly counter reset

🚀 Deployment Process

Deploying a New Chatbot

Step 1: Start Deployment

1. Verify prerequisites: All checks must be ✅
2. Click "Deploy Chatbot": Start the process
3. Confirm action: The process is not reversible during execution

Step 2: Deployment Phases

The system automatically performs these operations:

🔄 Starting Server (30-60 seconds)
   ├── Allocating server resources
   ├── Configuring environment
   └── Initializing services

🔄 Loading Knowledge Base (1-3 minutes)  
   ├── Uploading knowledge base
   ├── Indexing content
   └── Optimizing search

🔄 Update Models (2-4 minutes)
   ├── Loading AI models
   ├── Configuring NLP pipeline
   └── Automatic testing

✅ Deploy Completed!

Estimated total time: 3-8 minutes depending on the size of the knowledge base

Step 3: Completion

At the end, you will receive:

• ✅ Confirmation of deployment completed
• 🔗 Integration code for the website
• 🌐 Demo page URL for testing
• 📊 Analytics dashboard link

Update of an Existing Chatbot

When an Update is Necessary

Update Models (after changes to the knowledge base):

• New Q&As added
• Changes to existing Q&As
• Upload new documents
• Changes to the survey

Update Chatroom (after UI changes):

• Change of colors or fonts
• New settings configurations
• Changes to system messages
• Updates to privacy policy

Types of Updates

Quick Update (Chatroom Only)

Time: 30-90 seconds
Includes: 
├── Updating appearance
├── New UI settings
└── Chatroom configurations

Complete Update (Models + Chatroom)

Time: 2-5 minutes
Includes:
├── Retraining AI models
├── Updating knowledge base  
├── New Q&As and content
└── All UI changes

Update with Server Restart

Time: 3-8 minutes
Includes:
├── Complete server restart
├── System version update
├── Architectural changes
└── Major configuration updates

🔗 Integration into the Website

Integration Code

Standard Code

At the end of the deployment, you will receive code similar to this:

        <!-- BEGIN CRAFTER CHATROOM -->
        <script async src="YOUR_LINK"></script>
        <!-- END CRAFTER CHATROOM -->

Where to Insert the Code

✅ Correct Placement:

<!DOCTYPE html>
<html>
<head>
    <meta charset="UTF-8">
    <title>Your Site</title>
    
    <!-- Other head tags -->
    
    <!-- INSERT CHATBOT CODE HERE -->
    <script>/* Crafter.ai Code */</script>
    
</head>
<body>
    <!-- Page content -->
</body>
</html>

❌ Incorrect Placements:

• In the <body> → May cause loading delays
• After other scripts → Possible conflicts
• In external JS files → Timing issues

Integration Troubleshooting

Widget Not Appearing

Possible causes:

🔍 Check 1: Browser Console
   → Open Developer Tools (F12)
   → Look for JavaScript errors
   → Verify script loading

🔍 Check 2: Network tab
   → Verify download of bundle.js
   → Check response status (should be 200)
   → Ensure no CORS blocks

🔍 Check 3: Bot Status
   → Go to the Crafter.ai dashboard
   → Verify status = "Deployed"
   → Check last successful deploy

Common solutions:

🔧 CORS Error:
   → Verify authorized domain in dashboard
   → Add domain in bot settings

🔧 Script Blocked:
   → Check ad-blocker
   → Verify Content Security Policy
   → Whitelist cdn.crafter.ai

🔧 Incorrect Bot ID:
   → Copy/paste bot ID from dashboard
   → Ensure no extra spaces

Widget Appears but Does Not Work

Debug steps:

1. Test on official demo page
2. Compare behavior
3. Check widget version (latest vs specific)
4. Test in incognito mode
5. Disable other extensions/scripts

📊 Monitoring and Analytics

Deployment Dashboard

Real-Time Server Status

🟢 Deployed: Server active and functioning
🟡 Updating: Update in progress
🔴 Offline: Server offline or error
🟠 Outdated: Needs update

Logs and Diagnostics

Accessing logs:

1. Deployment Dashboard → "View Logs"
2. Filter by date/time
3. Levels: Info, Warning, Error
4. Export for in-depth analysis

Advanced Analytics

Performance Metrics

📈 Success Rate: % of conversations successfully resolved
📈 User Satisfaction: Average user rating (1-5 stars)
📈 Average Session Duration: Average conversation duration
📈 Bounce Rate: % of users who leave immediately
📈 Conversion Rate: % of completed goals

Behavioral Analytics

🔍 Most Asked Questions: Most frequent questions
🔍 Unresolved Queries: Questions without answers
🔍 User Flow Paths: Common conversation paths
🔍 Drop-off Points: Where users abandon
🔍 Peak Usage Times: Times of highest usage

Automated Reports

Configuring reports:

• Frequency: Daily, weekly, monthly
• Recipients: Team members via email
• Format: PDF executive summary or detailed Excel
• Custom metrics: Choose relevant KPIs

Alerting and Notifications

Automatic Alerts

⚠️ Downtime Alert: Server offline > 2 minutes
⚠️ High Error Rate: > 10% errors in 1 hour
⚠️ Low Success Rate: < 70% conversations resolved
⚠️ Usage Spike: Traffic > 300% normal
⚠️ Storage Limit: Knowledge base > 90% limit

Notification Channels

• Email: Immediate alerts via email
• Slack/Teams: Integration with corporate chat
• Webhook: Custom system integration
• SMS: Critical alerts (optional)

🔄 Managing Updates

Update Scheduling

Recommended Update Cycle

📅 Routine Updates:
├── Daily: Review new conversations
├── Weekly: Analyze misses and improvements
├── Monthly: Review performance and optimizations
└── Quarterly: Strategic review of knowledge base

🔄 Reactive Updates:
├── Immediate: Critical errors or incorrect information
├── Within 24h: New FAQs or frequent questions
├── Within a week: Significant UX improvements
└── On request: New business features

Best Practice Timing

Recommended times for updates:

• 🌙 Night hours: 02:00-05:00 (low traffic)
• 📅 Weekend: Saturday evening or Sunday morning
• 🚫 Avoid: Peak hours, promotional days

Hotfix Deployment

For urgent fixes:

1. Identify critical problem
2. Quick fix and essential testing
3. Immediate deployment with intensive monitoring
4. Communication with team and stakeholders
5. Post-mortem and lessons learned

🛠️ Preventive Maintenance

Maintenance Routines

Daily Checks (5-10 minutes)

✅ Server Status: Check uptime and performance
✅ Error Logs: Review errors from the last 24h
✅ New Conversations: Analyze new conversations
✅ Usage Metrics: Check resource usage

Weekly Checks (30-45 minutes)

📊 Analytics Review: Weekly performance analysis
🔍 Misses Analysis: Review unanswered questions
🔧 Knowledge Base Update: Content update
📱 User Feedback: Review ratings and comments
🧪 Regression Testing: Test critical functionalities

Monthly Checks (2-3 hours)

📈 Performance Report: Complete monthly report
🗂️ Knowledge Base Cleanup: Remove obsolete content
🔄 Workflow Optimization: Optimize complex flows
👥 Team Training: Update team on new features
💾 Backup Verification: Verify backups and recovery

Quarterly Checks (Half a day)

🎯 Strategic Review: Align business objectives
🔍 Comprehensive Audit: Complete system audit
🔧 Infrastructure Review: Review architecture and scalability
📚 Training Data Review: Clean and optimize data
🛡️ Security Audit: Review security and compliance

Performance Optimization

Monitoring Key KPIs

🎯 Target Performance:
├── Response Time: < 2 seconds
├── Uptime: > 99.5%
├── Success Rate: > 85%
├── User Satisfaction: > 4.0/5.0
└── Error Rate: < 5%

Common Optimizations

⚡ Knowledge Base:
├── Removal of duplicates
├── Consolidation of similar Q&As
├── Optimization of indexing
└── Cleaning of obsolete content

🚀 Server Performance:
├── Cache optimization
├── Database query optimization
├── CDN configuration
└── Resource scaling

📋 Common Troubleshooting

Deployment Issues

Deployment Fails

Error: "Deploy failed - server error"

Solutions:
1. Retry after 5-10 minutes
2. Check system status (status.crafter.ai)
3. Ensure plan limits are not exceeded
4. Contact support with error logs

Deployment Very Slow

Problem: Deployment takes > 15 minutes

Possible causes:
├── Very large knowledge base (> 100MB)
├── Many Q&As (> 500)
├── Temporary server overload
└── Network issues

Solutions:
├── Optimize knowledge base
├── Split deployment into multiple phases
├── Retry during low traffic hours
└── Check internet connection

Post-Deployment Issues

Chatbot Not Responding

Debug checklist:
1. Verify bot status = "Deployed"
2. Test on official demo page
3. Check integration code is correct
4. Verify authorized domain
5. Check browser console for errors

Incorrect Responses

Analysis:
1. Review conversation logs
2. Identify common error patterns
3. Verify knowledge base is updated
4. Check training completed correctly
5. Analyze Q&A conflicts

Degraded Performance

Investigation:
1. Monitor resource usage
2. Analyze response times
3. Verify database performance
4. Check concurrent user limits
5. Review error rate trends

✅ Deployment and Maintenance Checklist

Pre-Deployment Checklist

• All prerequisites verified
• Complete testing performed
• Team notified of deployment
• Backup configuration saved
• Rollback plan prepared
• Monitoring configured

Post-Deployment Checklist

• Deployment completed successfully
• Integration code implemented
• Widget functioning correctly
• Analytics active and configured
• Team training completed
• Documentation updated

Maintenance Checklist

• Routine checks implemented
• Alerts and notifications configured
• Performance baseline established
• Backup and recovery tested
• Team responsibilities assigned
• Escalation procedures defined

🔗 Next Steps

After successful deployment:

1. [Analytics and Optimization]
2. [Team Training]
3. [Advanced Features]

---

⏱️ Estimated first deployment time: 15-30 minutes (including integration) 🎯 Objective: Live, stable, and monitored chatbot in production