Everything your AI needs to understand your data

A deep look at each capability — from metrics definitions to visual lineage to AI verification.

Core Feature

Metrics Dictionary

Define every metric once — name, formula, definition, gotchas, and example queries — all in one place. When someone asks AI "what's our churn rate?", it uses your exact definition, not a guess.

  • Formula, description, and business context per metric
  • Link metrics to the tables and columns they depend on
  • Categorize by domain (finance, product, marketing)
  • AI-assisted metric creation from your existing data
Example Metric
name: "Monthly Recurring Revenue"
abbrev: "MRR"
formula: "SUM(amount_cents / 100) WHERE status = 'active'"
table: "subscriptions"
gotchas:
- "Exclude trial accounts (is_trial = true)"
- "Amount is stored in cents, divide by 100"
stored_in_cents

Column stores monetary values in cents. Divide by 100 to get dollars before displaying or aggregating.

soft_delete

Rows are not physically deleted. Filter by deleted_at IS NULL to exclude soft-deleted records.

test_data_present

Table contains test accounts. Filter by is_test = false for production data.

Data Safety

Gotcha Library

Every data warehouse has landmines — columns that look straightforward but aren't. Gotchas are reusable warnings that you tag to columns so AI handles them correctly every single time.

  • Reusable across multiple tables and columns
  • Severity levels: info, warning, critical
  • Automatically injected into AI context
  • AI suggests gotchas based on column names and types
Institutional Knowledge

Business Context

Your data doesn't exist in a vacuum. Contextary lets you document company-wide context — fiscal year boundaries, naming conventions, source system quirks, and the decisions that shaped your data model.

  • Company info, industry, and fiscal calendar
  • Naming conventions and coding standards
  • Source system documentation (CRM, billing, etc.)
  • Always included in AI context for accurate answers
Company Context
fiscal_year_start: "February 1"
default_currency: "USD"
id_convention: "UUIDs, never auto-increment"
Source Systems
crm: "Salesforce" — syncs hourly
billing: "Stripe" — amounts in cents
product: "Segment + Snowflake"
Data Lineage Graph
raw.stripe_charges
raw.salesforce_opps
staging.orders
analytics.revenue
analytics.orders_daily
Visibility

Visual Data Lineage

See where your data comes from and what depends on it. An interactive graph lets you trace any table upstream to its source or downstream to every dashboard that depends on it.

  • Interactive DAG visualization
  • Trace upstream sources for any table
  • See downstream impact before making changes
  • Import lineage from dbt or define manually
Quality Assurance

AI Verification

Don't just trust the answer — verify it. Ask Contextary a question in plain English, see the query it generates, run it against your real data, and confirm the results make sense before sharing them.

  • Natural language to SQL, grounded in your context
  • Run queries against your actual warehouse
  • See which context was used to generate the query
  • Thumbs up/down feedback to improve over time
Ask Data
Your question
"How many active users last month?"
Generated SQL
SELECT COUNT(DISTINCT account_uuid)
FROM events
WHERE event_date >= '2026-03-01'
AND deleted_at IS NULL
AND is_test = false
Result
8,247 active users
JK
Julia K.
Updated metric: "Net Revenue Retention" — 2 hours ago
AM
Alex M.
Added gotcha: "soft_delete" to 4 tables — yesterday
RP
Ravi P.
Annotated 12 columns on "orders" — 3 days ago
Collaboration

Team Collaboration

Data knowledge shouldn't live in one person's head. Your whole team can contribute annotations, metrics, and gotchas — building a shared understanding that compounds over time.

  • Activity feed shows who changed what
  • Knowledge grows with every annotation
  • Onboard new team members faster
  • Roles and permissions (Team plan)
Integration

Warehouse Sync

Push your documented context back into your warehouse as table and column comments. Your BI tools — Looker, Tableau, dbt — pick up the descriptions automatically.

  • Sync descriptions to warehouse metadata
  • BI tools show context without extra setup
  • BigQuery, Snowflake, Postgres, Redshift
  • One-click sync or automated schedule
BigQuery
Snowflake
Postgres
Redshift
Compatibility

Works with every AI tool you use

Contextary serves your data knowledge over the Model Context Protocol (MCP) — an open standard supported by Claude, Cursor, Claude Code, and a growing list of AI tools. Your context is always available, no matter which tool your team prefers.

Claude Desktop
Chat with your data
Cursor
Context-aware coding
Claude Code
Terminal-first analytics
Any MCP Client
Open protocol

Ready to give your AI the full picture?

Free to get started. Ready in minutes.