How Gorgias Onboarding Works Today
Sequential process: Setup everything first, then AI starts learning
Connect Channels
Shopify, Email, Chat
Day 1
Build Help Center
Articles, FAQs
Day 2-5
Write Guidance
Policies, Processes
Day 5-10
Configure Actions
Shopify, Returns
Day 10-14
Enable AI Agent
Start from scratch
Day 14+
The Problem
During the entire setup period (often 2+ weeks), human agents are already handling tickets—learning patterns, developing tone, figuring out what works. But the AI isn't watching any of this. When it finally turns on, it starts cold with zero context from your team's actual experience.
Proposed: Watch & Learn
AI learns passively during setup, launches calibrated, improves continuously
👀 Parallel Learning
AI observes while you onboard
- ✓ Watches how agents handle tickets
- ✓ Learns tone, patterns, edge cases
- ✓ Identifies common ticket types
- ✓ Zero extra work for your team
During Setup
👨💼
Human Agents
Handling tickets
🤖
AI Agent
Watching & learning
📊 AI learns: Tone patterns, common queries, resolution approaches
🔬 Beta Comparison
AI + Humans handle same tickets
- ✓ Side-by-side resolution comparison
- ✓ Automatic gap identification
- ✓ KB suggestions generated
- ✓ Confidence scoring per topic
Beta Testing Mode
👨💼
Human
4.6 ⭐
⚔️
🤖
AI
4.2 ⭐
💡 Gap detected: AI needs empathy templates for WISMO queries
🚀 L2 Expansion
Gradually automate complex tickets
- ✓ Move from L1 → L2 automation
- ✓ Watch returns, billing, exceptions
- ✓ Same loop, higher complexity
- ✓ Human stays on true edge cases
Automation Expansion
L1 - Basic
85% automated
WISMO, Returns policy, Sizing
L2 - Complex
45% automated
Exceptions, Billing, Complaints
L3 - Edge Cases
10% automated
Escalations, VIP, Legal
⚡
Faster Time-to-Value
AI is calibrated by Day 1, not starting cold after weeks of setup
🔄
Passive Feedback Loop
Every handover becomes training data. No manual labeling needed.
🎯
Zero Extra Work
Brand is already doing setup + handling tickets. Just capture it.
AI vs Human: Beta Comparison
Side-by-side performance on same ticket types. Gaps become KB suggestions.
Tickets Compared
699
AI Win Rate
40%
Gaps Identified
3
Action needed
Avg CSAT Delta
-0.3
Improving
| Ticket Type | Volume | AI Time | Human Time | AI CSAT | Human CSAT | Winner |
|---|---|---|---|---|---|---|
| Where is my order?GAP | 342 | 0.8s | 4m 12s | 4.2 | 4.6 | 👨💼 Human |
| Return request | 156 | 1.2s | 6m 45s | 4.5 | 4.4 | 🤖 AI |
| Product sizing helpGAP | 89 | 2.1s | 3m 22s | 3.1 | 4.8 | 👨💼 Human |
| Discount code issues | 67 | 0.9s | 2m 15s | 4.7 | 4.3 | 🤖 AI |
| Damaged item receivedGAP | 45 | 1.5s | 8m 30s | 3.4 | 4.9 | 👨💼 Human |
3 Knowledge Base Gaps Identified
Click to see auto-generated suggestions
▼
● Where is my order?
Missing warmth in AI response. Suggest: Add empathy templates to KB.
● Product sizing help
AI lacks product-specific knowledge. Suggest: Add sizing guide + fit tips to KB.
● Damaged item received
Sensitive topic needs human touch. Suggest: Add to handover topics + create empathy script.