Physical data elements are extracted from your documents by specialized AI agents, or added manually by your team
AI-powered extraction
The agent parses database schemas, DDL scripts, and data dictionaries to extract column definitions with full metadata including data types and constraints.
Processes these document types:
Schema Parser
Understands DDL, Excel schemas, and technical docs
Manual entry & review
Database administrators and data engineers can add columns directly or validate and approve agent-extracted schemas.
Human actions:
Full Control
Every AI suggestion requires human approval
Every record carries a trust score based on its source document maturity level. High-maturity documents (like approved Technical Specifications) produce higher-trust inventory records.
High Trust
From production DDL scripts, approved schemas
Medium Trust
From draft schemas, UAT environments
Low Trust
From design docs, informal sources
Trust Score Distribution
Records start as "pending" and must be approved before they are used in downstream processing like lineage generation
Pending
Awaiting review
Use
Ready for lineage
Lineage
Connected to pipelines
Setting a physical element to "Use" enables it for semantic matching — IDA will automatically find business terms that describe this database column.
Catalog database columns with full technical metadata including schemas, data types, and constraints
| # | Trust | Action | Column Name | Table Name | Schema | Data Type | System | Source Documentation |
|---|---|---|---|---|---|---|---|---|
| 1 | High | Use | CUST_ID | CUSTOMERS | CRM | VARCHAR(50) | CRM System | CRM Database Schema v4.xlsx |
| 2 | High | Use | ACCT_BAL | ACCOUNTS | FINANCE | DECIMAL(18,2) | ERP Platform | ERP Data Dictionary.xlsx |
| 3 | Medium | Use | TXN_DATE | TRANSACTIONS | PAYMENTS | DATETIME | Payment Gateway | Payment Schema Documentation.pdf |
| 4 | Medium | Pending | PROD_NM | PRODUCTS | CATALOG | VARCHAR(200) | Product Catalog | Product Database DDL.sql |
| 5 | Low | Pending | RISK_SCORE | CREDIT_ASSESSMENTS | RISK | INTEGER | Risk Engine | Risk Model Technical Spec.docx |
| 6 | Low | Pending | ADDRESS_LINE_1 | CUSTOMER_ADDRESSES | CRM | VARCHAR(255) | CRM System | Manual entry via UI |
This is a preview — the real interface supports full CRUD operations
Physical elements represent implementation — explore all four inventories that form your complete data landscape
See how IDA can automatically extract column definitions from your DDL scripts and data dictionaries in a 30-minute demo.