How ZLens fits with dbt

dbt is the modeling tool. ZLens is the reporting + governance layer. They share a single source of truth: target/manifest.json.

Division of concerns

Concern	dbt	ZLens
Transform raw → marts	✅	—
Tests on models	✅	—
Semantic layer (MetricFlow)	✅ authoring	✅ import + extend
Lineage at the model level	✅	✅ (absorbs dbt's)
Lineage at the block / dashboard level	—	✅
Notebook authoring	—	✅
Certified reusable analytics blocks	—	✅
Static HTML dashboards	—	✅
Governance (owner, certification, lint)	partial	✅
BI-style exploration	—	✅

The integration surface

ZLens reads target/manifest.json. That's it — no dbt-API calls, no YAML duplication, no "ZLens's version of a model." When present, ZLens also reads target/semantic_manifest.json for MetricFlow semantic models, metrics, measures, dimensions, entities, and saved queries.

dbt models     ─▶  target/manifest.json  ─▶  ZLens semantic layer
dbt metrics    ─┘                         │
dbt sources    ─┘                         ▼
                                     DQL blocks
                                     DQL notebooks
                                     DQL dashboards

Workflow

# In your dbt project:
dbt build

# In your ZLens project:
zlens compile .
zlens sync dbt .
zlens agent reindex

Add these ZLens commands to your CI after dbt build: zlens compile . rebuilds the DQL manifest and lineage graph, zlens sync dbt . verifies artifact paths and cache status, and zlens agent reindex refreshes the governed answer index.

What ZLens does not do

• Transform raw data. Use dbt for that.
• Schedule model runs. Use dbt + Airflow/Dagster/dbt Cloud.
• Replace MetricFlow. ZLens extends it, never competes with it.

What to move to ZLens

The user-facing outputs of your analytics stack: reports, dashboards, ad hoc notebooks, certified shared blocks. All the stuff that used to live in Metabase/Looker/Hex and had no good git story.