See How CX-Catalyst Transforms Support
From submission to resolution, understand exactly how AI handles every support request.
The Journey of a Support Request
Request Arrives
A customer submits a support request through any channel:
- Web form
- Chat widget
- API integration
The request includes their issue description, any attachments, and customer context.
Smart Intake & Classification
Within milliseconds, AI analyzes the request:
Category: Authentication
Priority: Medium
Sentiment: Frustrated
Confidence: 87%
Routing: Self-Service
Intelligent Routing
Based on confidence and priority, the request is routed:
| Confidence | Route |
|---|---|
| 85%+ | Self-Service Resolution |
| 60-85% | Collaborative Review |
| Below 60% | Human Escalation |
Resolution Path
The appropriate workflow handles the request:
- Self-Service (85%) – AI searches KB, generates personalized response, sends to customer
- Collaborative (10%) – AI drafts response, team reviews via Slack, then sends
- Escalation (5%) – Assigned to agent with full AI-provided context
Customer Resolution
Customer receives their response:
- Self-Service: Average 2-5 minutes
- Collaborative: Average 15-30 minutes
- Escalation: Based on priority SLA
Feedback Loop
After resolution:
- Customer can mark “Helpful” or “Not Helpful”
- Negative feedback triggers review
- All interactions feed into learning system
- Knowledge base improved continuously
Resolution Paths Explained
Self-Service (85% of requests)
When it’s used:
- AI is confident it understands the issue
- Similar issues have been successfully resolved before
- Knowledge base has relevant documentation
Example: “How do I reset my password?”
AI recognizes authentication category, finds password reset KB article, generates step-by-step instructions. Customer receives response in 2 minutes.
Collaborative (10% of requests)
When it’s used:
- AI understands but isn’t fully confident
- Issue requires verification before sending
- Sensitive categories (billing, account changes)
Example: “I was charged twice for my subscription”
AI identifies billing issue, drafts refund process response, sends to billing team for review. Agent verifies and approves with edits.
Human Escalation (5% of requests)
When it’s used:
- AI cannot confidently classify the issue
- Complex multi-part requests
- Unique situations not in knowledge base
Example: “My entire account data disappeared after the update”
AI recognizes potential data loss (critical), confidence too low for auto-response. Escalates to senior support immediately with full context.
AI Decision Making
How Classification Works
The AI uses advanced natural language processing to understand:
- Semantic Understanding – Not just keywords, but meaning and intent
- Handles variations – Typos, slang, different terminology
- Context awareness – Understands based on customer history
Confidence Scoring
Confidence represents AI’s certainty in its understanding:
High (85%+)
Clear, well-defined issue. Strong match to known patterns. Unambiguous intent.
Medium (60-85%)
Issue understood but some uncertainty. Multiple possible interpretations.
Low (<60%)
Unclear or complex issue. No matching patterns. Contradictory signals.
See It in Action
Watch CX-Catalyst process real support requests in our interactive demo.