01 Lineage

Lineage you didn't have to draw.

EVE builds column-level lineage automatically from dbt runs and executed SQL. It is derived from what actually ran, not maintained by hand in a diagram that goes stale.

  • Automatic, not hand-drawn. Every dbt run and executed query updates the graph, so lineage matches reality without anyone maintaining it.
  • Column-level, end to end. Follow a single column from raw source through staging and marts to the dashboard that shows it.
  • Blast-radius view. Before you change a column, see everything downstream that depends on it.
  • Quality-failure overlay. A failing test lights up every asset it touches, so impact is obvious at a glance.
lineage · column-level
Column-level lineageColumn-level lineage from raw orders through staging and a mart to a dashboard, with the amount to revenue column edge highlighted.ordersorder_idemailamountstg_ordersorder_idemailamountcustomer_360customeremailltvrevenue_dashrevenueregion
Automatic from dbt runs and executed SQL. The amount → revenue edge is a real column dependency, not hand-drawn.
02 Data quality

Quality with an ops workflow.

Author tests without code, run them natively where the data lives, and manage failures as incidents with a real lifecycle - not a wall of red that nobody owns.

AUTHOR

Point and click

Build tests visually, no code required, and group them into suites that run together on a schedule.

not_null · unique · accepted_values
RUN

24 types, 6 dialects

24 test types run natively against 6 SQL dialects, right where the data lives - no extra engine to stand up.

24 types · 6 dialects
RESOLVE

Incident lifecycle

New to Ack to Assigned to Resolved, with auto-reopen if it starts failing again. Result history is charted over time.

New > Ack > Assigned > Resolved
03 PII masking

PII stays private - by default.

Two layers, not one setting. Sensitive columns are masked in the data you see, and the profiler refuses to compute statistics on them - so PII cannot leak through the back door of a summary stat.

  • Layer one - masked in the data. Previews, samples, and query results show masked values instead of raw PII, for anyone without clearance.
  • Layer two - never profiled. The profiler never computes statistics on PII-tagged columns, so nothing leaks through min, max, or distinct counts.
  • Auto-classification. Pattern and deny-list detectors tag PII columns automatically, so protection does not wait on someone remembering to flag them.
  • Enforced everywhere. Queries, statistics, and AI answers all respect the same tags - one policy, every surface.
customers · column policy
What a viewer without clearance sees
email · PII
j••••@•••.com
masked in samples and results
ssn · PII
•••-••-••••
masked in samples and results
amount · not PII
1,204.00
shown - no sensitive tag
Profiler
statistics skipped on 2 PII columns
no stats computed on tagged columns
04 Vocabulary and contracts

One vocabulary for the whole estate.

Terms, tags, metrics, and contracts live in the catalog with everything else - so what a number means, and who may see it, is defined once.

Glossary & tags

One shared vocabulary

Define business terms and tags once and attach them across the catalog. Import and export the whole glossary as CSV.

Metrics

Define a metric once

A metrics registry links each definition to its backing tables through lineage, so the same number agrees in every dashboard.

Contracts

Data contracts

Run contracts in validate-only mode on a schedule to flag a breach before it reaches a consumer.

Domains

Enforced, not filtered

Domains restrict access org-wide across every environment and product - a real boundary, not a cosmetic filter chip.

05 In every tier

Governance in every tier.

PII masking, column-level lineage, quality incidents, contracts, and domains ship in every plan. Never behind an edition paywall.

  • PII masking
  • lineage
  • quality
  • contracts
  • domains
  • metrics
Get started

See your data on EVE.

A 30-minute demo on your use case, or a free workspace on your own isolated cluster.