Skip to main content
When you select a table on the Intelligence page, you see its Object Profile: an AI-generated analysis that combines schema metadata with context from freshness, data quality, schema drift, tags, and lineage into a single view. The profile is only as useful as the monitoring you’ve set up. A table with freshness schedules, data quality metrics, and schema drift monitoring will have a much richer profile than one with no monitoring at all.
What users see when they select a table on the Intelligence page

​
What’s in a Profile

​
Object Summary

The top card shows the table’s identity at a glance:
  • Name and path: warehouse.gold.fact_orders
  • Warehouse role: Classified as Fact Table, Dimension, Staging, Raw, or other patterns
  • Confidence score: How confident the AI is in its analysis (based on available context)
  • Summary: One-paragraph explanation of what this table is for
  • Business context: What team uses it, what process populates it, what it means to the business
  • Domain and tags: Business and technical classifications applied to this table

​
Findings

Findings are the most actionable part of the profile. They surface issues and observations pulled from across your monitoring domains, grouped by severity:
SeverityExample
Critical”Column order_total was removed on Tuesday. 3 downstream tables reference this column.”
High”Table is 6 hours overdue for update. SLA threshold is 2 hours.”
Medium”Null rate on email increased from 2.1% to 12.4% over the past week.”
Low”Table has no freshness monitoring configured.”
Each finding includes the affected object, a description, and the monitoring domain it came from. Critical and High findings are expanded by default. Low findings are collapsed.

​
Relationships and Lineage

Shows how this table connects to the rest of your data:
  • Foreign key relationships: Detected column references to other tables
  • Lineage signals: Whether this table is a source, derived, or staging table
  • Schema-level lineage: Which schemas act as sources vs. consumers
  • Hub tables: Central entities that many tables reference

​
Change History

A timeline of recent schema changes detected on this table:
  • Column additions (green)
  • Column removals (red)
  • Type modifications (amber)
Shows the 10 most recent changes with timestamps. This context is what lets Intelligence answer questions like “what changed on the orders table this week?”

​
Analysis Context

Shows which monitoring domains have data available for this table. The agent uses this context when answering questions:
DomainWhat It Provides
SchemaTable structure, column types, constraints
TagsBusiness classifications, PII labels
FreshnessUpdate patterns, SLA status, learned schedules
RulesAlert rule configurations and firing history
Data QualityMetric trends, validity rule results
A checkmark means context is available. Missing domains mean you haven’t set up that monitoring yet.

​
Generating Profiles

​
Single Table

  1. Navigate to a table in the Schema Explorer
  2. Click Analyze Table in the top right
  3. Wait for analysis to complete (typically 10-30 seconds)
  4. The profile appears with all available context

​
Entire Schema

  1. Select a schema in the Schema Explorer
  2. Click Analyze Schema
  3. All tables in the schema are analyzed

​
Full Database

  1. Select the database root in the Schema Explorer
  2. Click Analyze All
  3. Every table across all schemas is analyzed
Analysis runs in the background. You’ll see a notification when it completes, and the profiles update automatically.

​
Making Profiles More Useful

The quality of a profile depends directly on what monitoring you’ve set up:
What You Set UpWhat the Profile Gains
Freshness scheduleUpdate patterns, SLA violation findings, staleness history
Schema drift monitoringChange timeline, removed column warnings, drift findings
Data quality metricsTrend data, anomaly findings, null rate tracking
TagsBusiness context, PII classification, domain grouping
Alert rulesAlert coverage analysis, firing history context
Start with your most critical tables. Set up freshness monitoring and a few data quality metrics, then generate intelligence. The profile will immediately surface findings from that context.

​
Editing Descriptions

AI-generated descriptions are a starting point. You can edit any description to add context the AI can’t infer:
  • Business context: “Used by the finance team for quarterly reporting”
  • Data sources: “Populated by the Stripe webhook integration”
  • Update frequency: “Updated in real-time as orders are placed”
  • Caveats: “Does not include cancelled orders before 2023”
Edits are preserved across re-analysis. Your additions won’t be overwritten.

​
Next Steps

Ask Questions

Use profiles as context for AI-powered Q&A

Walkthrough

See profiles in action during a real debugging scenario