Skip to main content

Statistical Profiling

ekai performs full-column statistical profiling on your data—no sampling. This analysis runs in-place within your warehouse and powers the AI's key detection algorithms.

Prerequisites

Complete the Onboarding Agent interview to provide business context.

Profiling Overview​

After onboarding completes, ekai automatically profiles all selected tables:

Profiling completed view

Profile Tab Navigation​

The PROFILE tab shows:

ElementDescription
Total RowsSum of all rows across profiled tables
Table TabsNavigate between tables
Usage ContextRow count, data volume, last modified
Column AnalysisDetailed statistics per column

Table-Level Metrics​

For each table, ekai displays:

Row Count

Total rows in the table

Data Volume

Approximate size in MB/GB

Last Modified

When data was last updated

Column-Level Analysis​

Click on any column to see detailed statistics:

What Gets Analyzed​

ekai analyzes schema metadata and statistical properties to understand your data structure:

  • Data types — Column types and formats
  • Schema patterns — Table and column naming conventions
  • Statistical properties — Aggregated characteristics for AI analysis
  • Relationships — Connections between tables

All analysis runs in your warehouse. Only statistical summaries are transferred—never raw data.

How Profiling Powers Key Detection​

ekai uses AI to analyze your schema and data patterns, combining multiple signals to accurately detect primary and foreign keys with confidence scores.

Re-Running Profiles​

To refresh statistics after data changes:

  1. Click Restart Profiling button (top right)
  2. Wait for profiling to complete
  3. New statistics are used for ERD generation
Restart Profiling

Privacy & Security​

AspectGuarantee
Data LocationQueries run in your warehouse
Data TransferOnly statistics transferred, not raw data
StorageStatistical features stored, not source data
EncryptionTLS 1.3 for all transfers

Next Steps​