Generate ERD

ekai generates an Entity Relationship Diagram with AI-detected primary and foreign keys using multi-signal analysis from statistical profiling and onboarding context.

Prerequisites

Complete Statistical Profiling and the Onboarding Agent.

---

Starting ERD Generation

Click Go To Schema Generation or navigate to the SCHEMA tab to start:

The AI Agent initiates schema analysis with connection information:

• Connection name and type
• Total number of tables
• Processing status

---

Generation Progress

Watch the AI work through your schema:

Progress Steps

The AI Agent Chat shows real-time progress:

To Do:

• Take inventory (profile + onboarding doc) ✓
• Filter FK/PK candidates from profile ✓
• Run schema workflow (keys, relations, docs) ✓
• Commit schema to memory
• Independent review against profile
• Apply agreed corrections
• Present schema to user

---

Key Detection Results

For each table, the SCHEMA tab displays detected keys:

Keys Tab

Field	Description
Primary Key	Detected primary key column(s)
Type	Single or Composite
Foreign Keys	Detected FK columns
Confidence	Certain, High, Medium, Low
Reasoning	AI explanation for the detection

Example Detection

Table: SALESFORCE_CAMPAIGNS
─────────────────────────────
Primary Key:
  • Columns: CAMPAIGN_ID
  • Type: Single
  • Confidence: Certain

Foreign Keys:
  • PRODUCT_ID → SALESFORCE_PRODUCTS.PRODUCT_ID

Analysis:
  "CAMPAIGN_ID is the clear candidate for the primary key as it
  uniquely identifies a campaign. PRODUCT_ID shows a strong
  relationship (0.90) to SALESFORCE_PRODUCTS with an exact
  column name match."

---

ERD Complete

When generation completes, view the ERD and agent summary:

DBML Output

Table SALESFORCE_CAMPAIGNS {
  ACTUAL_SPEND float [note: "Actual expenditure for the campaign"]
  CAMPAIGN_ID number [pk, note: "Unique identifier for each campaign"]
  CREATED_DATE text [note: "When the campaign was created"]
  PRODUCT_ID number [note: "Related product identifier"]
  STATUS text [note: "Campaign status (active, inactive, completed)"]
  // ... more columns

  Note: "Marketing campaigns managed through Salesforce"
}

Table SALESFORCE_PRODUCTS {
  PRODUCT_ID number [pk]
  PRODUCT_NAME text
  // ... more columns
}

Ref: SALESFORCE_CAMPAIGNS.PRODUCT_ID > SALESFORCE_PRODUCTS.PRODUCT_ID

SCHEMA Sub-tabs

Tab	Content
KEYS	Primary and foreign key analysis per table
RELATIONSHIPS	All detected relationships
DOCUMENTATION	Generated table/column descriptions
ENTITY RELATIONSHIP	DBML code view

---

Visual ERD Diagram

Click View Visualization or open the ERD canvas for the interactive diagram.

Diagram Features

Interactive Canvas

• Zoom in/out
• Pan to navigate
• Click tables for details

Table Boxes

• Table name header
• Column names listed
• PK indicator

Relationship Lines

• FK connections
• Cardinality indicators
• Click for details

Controls

• + / - zoom
• Fullscreen mode
• Download image

---

Generation Summary

The AI Agent provides a summary of what was generated:

Database structure for your financial services platform:

26 tables across 3 systems:

• Salesforce (5 tables): Campaigns, interactions, leads, opportunities, products
• Temenos Core Banking (16 tables): Customers, accounts, loans, payments, KYC
• Worldline Payments (5 tables): Cards, transactions, merchants, disputes, fraud alerts

Key highlights:

• 33 relationships connecting your data
• Primary and foreign keys identified across all tables
• Detailed descriptions for every table and column
• Full integration between CRM, banking, and payment systems

---

Next Steps

• Review Relationships — Refine ERD with AI chat
• Semantic Models Overview — Use ERD to build data products