Screen 1 — Agent Selection

Your Agents

TAZI Platform / Agents
📉
Churn Prediction
Upload customer data. Get churn risk scores and actionable retention strategies.
Ready
🚨
Fraud Detection
Upload transaction data. Get flagged anomalies with explanations.
Coming soon
📈
Demand Forecasting
Upload sales data. Get forecasts with confidence intervals.
Coming soon
Screen 2a — Upload & Configure (Step 1 of 4)
Agents / Churn Prediction
1
Upload Data
2
Data Quality
3
Processing
4
Results
Drop your customer data file here, or click to browse
CSV or Excel • Max 500MB • One row per customer
Which column indicates whether a customer churned?
Select the column that contains "churned" / "active" or "1" / "0"
What is your primary business KPI for retention?
This helps the agent focus on the most relevant insights
Prediction window
How far ahead should the model predict churn risk?
Screen 2b — Data Quality Report (Step 2 of 4)
Agents / Churn Prediction
✓
Upload Data
2
Data Quality
3
Processing
4
Results

We analyzed 12,847 rows across 24 columns. Here's what we found:

Overall Data Quality 2 issues
Rows: 12,847 • Columns: 24 • Target: churn_status (binary)
Class balance: 78% active, 22% churned — acceptable
monthly_spend OK
Type: numeric • Missing: 0.2% • Range: $4.99 – $899.00
last_login_date Needs attention
Type: date • Missing: 12.3%
⚠ 12.3% of rows have no last login date. This is a strong churn predictor. Consider filling missing values or the model may underpredict risk for these customers.
region Needs attention
Type: categorical • Unique: 147 • Missing: 0%
⚠ 147 unique regions is very high cardinality. Consider grouping into broader regions (e.g. country or continent) for better model performance.
You can re-upload corrected data or proceed as-is. The agent will handle missing values automatically.
Screen 2c — Processing (Step 3 of 4)
Agents / Churn Prediction
✓
Upload Data
✓
Data Quality
3
Processing
4
Results
Your churn prediction agent is working...
Training model on 12,847 customers • Estimated 3-5 minutes
You'll be notified when results are ready. You can close this page.
Screen 3 — Results & Conversation (Step 4 of 4)
Agents / Churn Prediction / Results
Churn Agent
Your churn model is ready. Here's the summary:

234 customers (1.8%) are at high risk of churning in the next 90 days, representing $47,200/mo in MRR at risk.

The top risk factors are: days since last login, support ticket frequency, and contract type.
Top churn risk factors (feature importance)
Which customers should I reach out to first?
Churn Agent
Here are the top 10 customers by combined risk score and MRR impact:

1. Acme Corp — $2,400/mo, 94% churn risk. Last login: 45 days ago. 3 open support tickets.
2. TechFlow Inc — $1,800/mo, 91% churn risk. Downgraded plan last month.
3. DataBridge — $1,200/mo, 88% churn risk. Zero logins in 60 days.

Want me to generate a retention action plan for these accounts?
Model Summary
Total customers12,847
High risk234 (1.8%)
MRR at risk$47,200
Model accuracy92.3%
Prediction window90 days
Ask about your results
Understanding risk
Which customers are most likely to churn?
What are the top risk factors driving churn?
Show me the churn risk distribution
Taking action
Generate a retention action plan
Which segments should I focus on?
What would reducing churn by 10% save us?
Going deeper
Why is Acme Corp flagged as high risk?
Compare churn rates by plan type
Export high-risk customer list