Onboarding Agent
The Onboarding Agent conducts an AI-driven interview to understand your data architecture. This context enables more accurate ERD generation by incorporating business knowledge alongside statistical analysis.
Complete the Create Connection step to connect to your data warehouse.
How It Works
When you open a Schema Agent connection, the Onboarding Agent:
- Analyzes your connected tables and columns
- Identifies potential source systems and data domains
- Asks targeted questions (maximum 5) to clarify business context
- Generates a summary of your data architecture
The AI Agent Chat
The right panel shows the AI Agent Chat where you interact with the Onboarding Agent:
Welcome Message
The agent introduces itself and explains what it will do:
"Welcome to EKAI! We specialize in generating accurate Entity Relationship Diagrams (ERDs) by intelligently analyzing your data ecosystem.
I see you have connected a Snowflake environment containing 26 tables.
To ensure your ERD reflects your true business logic, I'd like to ask a few targeted questions (maximum 5).
Shall we begin by analyzing your table structures?"
Chat Interface
- Messages — AI questions and your responses
- Input field — Type your answers
- Send button — Submit response
- Go To Schema Generation — Skip to ERD when ready
Example Conversation
Onboarding Agent:
"Based on the table naming patterns, I've identified what appears to be:
- Temenos tables (TEMENOS_*) — likely core banking
- Salesforce tables (SALESFORCE_*) — CRM data
- Worldline tables (WORLDLINE_*) — payment processing
Is this assessment correct? Are there other source systems I should know about?"
User:
"Yes, that's correct. Temenos is our core banking system, Salesforce handles CRM, and Worldline processes card payments. Customer ID is the primary link between systems."
Onboarding Agent:
"Thank you! That helps establish the integration pattern. One more question: For the TEMENOS_CUSTOMERS table, does CUSTOMER_ID serve as the master customer identifier across all systems?"
Onboarding Complete
When the interview is complete, the agent generates an ERD Onboarding Summary:
Summary Contents
The generated summary includes:
System Architecture Overview
Identified source systems and their roles in your data ecosystem.
Primary Systems
Each system with its function (e.g., Core Banking, CRM, Payments).
Architecture Pattern
The overall pattern (e.g., "Financial Services - Integrated Banking & CRM").
Data Integration
How systems connect (shared keys, data flow patterns).
Example Summary Output
## ERD Onboarding Summary
### System Architecture Overview
### Primary Systems Identified:
- **Temenos** - Core Banking System (handling Customers, Accounts, Loans, Beneficiaries)
- **Salesforce** - CRM (managing Leads, Opportunities, Campaigns)
- **Worldline** - Payment Processing (tracking Cards, Transactions, Fraud Alerts)
### Architecture Pattern
Financial Services (Integrated Banking & CRM)
### Data Integration
- Unified customer view established via shared Customer ID primary key
- Flow tracks lifecycle from acquisition (Salesforce) to account creation
(Temenos) and usage/payments (Worldline)
Tips for Better Results
| Tip | Why It Helps |
|---|---|
| Answer completely | More context = more accurate ERD |
| Mention integration keys | Helps identify foreign key relationships |
| Describe source systems | AI understands data patterns better |
| Share business rules | Custom rules get incorporated |
Mission Control Progress
After completing onboarding, Mission Control updates:
Next Steps
- Statistical Profiling — Analyze data patterns
- Generate ERD — Create the ERD