Skip to main content
These workflows work in both the in-app AI assistant and via MCP in your AI IDE.

​
Set Up a Cross-Sell Campaign

Goal: Create a complete cross-sell campaign from scratch using natural language. 
1. "Create a category called 'Cross-Sell' with description 'Product cross-selling campaigns'"

2. "Create an offer called 'Premium Credit Card' in the Cross-Sell category
    with priority 8 and daily budget of 1000"

3. "Create a qualification rule for Premium Credit Card:
    customer credit score must be above 700 and account age over 12 months"

4. "Create an email creative for Premium Credit Card with subject
    'Upgrade to Premium' and copy about exclusive rewards"

5. "Create a decision flow called 'Cross-Sell Flow' that includes
    the Premium Credit Card offer with propensity scoring"

​
Diagnose Declining Conversions

Goal: Figure out why conversion rates dropped. 
1. "Analyze offer performance for the last week"
   → See which offers have declining conversion rates

2. "Explain why customer C-5678 didn't receive the Home Loan offer"
   → Understand the specific funnel stage where customers are being filtered

3. "Analyze the qualification funnel for the Home Loan decision flow"
   → Find the biggest bottleneck rule

4. "Are there any policy conflicts affecting the Home Loan offers?"
   → Check for contradictions or over-aggressive suppression

​
Simulate a Policy Change

Goal: Test the impact of a policy change before applying it. 
1. "What is the current email frequency cap?"
   → Shows current contact policy settings

2. "Simulate what happens if I change the email frequency cap from 3 to 5 per week"
   → Shows estimated customer reach change and fatigue risk

3. "Simulate what happens if I lower the minimum credit score
    from 700 to 650 on the Premium Card qualification rule"
   → Shows before/after reach estimates

​
Build a Qualification Rule (Natural Language)

Goal: Create complex qualification rules without learning the rule syntax. 
"Create a qualification rule that targets customers who:
 - Have a credit score above 680
 - Have been a customer for at least 6 months
 - Have not received more than 2 offers this week
 - Are in the 'active' segment"
The AI assistant will parse this into the correct rule configuration and preview it for your approval.

​
Monitor Model Health

Goal: Check if your ML models need attention. 
1. "Analyze the health of all my models"
   → AUC, precision, recall, trends

2. "Is there any drift in the propensity model?"
   → Distribution comparison, calibration check

3. "What improvements would you suggest for the churn model?"
   → Missing predictors, hyperparameter recommendations