Single Control Plane for Data Management
One plane for context, lineage, governance, and secure data movement. Every agent works from the stack you already run, with proof attached.
Live demo · Walk through sign-in → connect → sync → dashboard · Setup in under 5 minutes
Multiple offerings.
One control plane.
Workspace for builders, Context Graph for understanding, Embedded Analytics for everyone else, unified on the same semantic foundation.
Where data teams build
Stack-aware editors, living catalog, and agents in one open workspace.
Explore Workspace →How your stack makes sense
One graph linking Snowflake, dbt, Airflow, BI, and business metrics.
Explore Context Graph →Metrics everyone trusts
Certified KPIs in Looker, board decks, and apps your org already uses.
Explore Embedded Analytics →Connect once. Understand everything, technical and business.
One intelligence layer: deep infrastructure graph plus canonical metrics your whole org can trust.
Living catalog that stays fresh
Schemas, lineage, ownership, and governance, discovered automatically from Snowflake, dbt manifests, and orchestration metadata. Never manually tagged again.
Infrastructure AI
Ask in plain English. Get answers grounded in your actual DAGs and models.
Stack-aware editors
SQL and dbt autocomplete that knows your refs, schemas, and lineage.
Interactive lineage
Click any node. Trace upstream and downstream impact in seconds.
Canonical metrics
Define KPIs once. Standardize across tools. Reconcile with full lineage back to source.
Company Brain
Infrastructure-aware AI plus business reasoning. Ask about lineage, impact, or why churn moved last week. Grounded answers with citations.
Built for your role.
Proven in your industry.
Concrete playbooks not generic "data workflows." Pick a persona below.
Pick a role or industry to see a playbook.
Ship dbt PRs without breaking production
Run impact analysis before every merge. Agents generate schema.yml tests from your manifest. PRs ship with lineage diffs attached.
- Slack threads for lineage
- Manual impact spreadsheets
- Surprise failures on merge
- Stack-aware editor + impact query
- Agent-generated tests
- Lineage diff on every PR
Debug DAG failures with full stack context
See daily_revenue linked to dbt models. Agents read retry logs. Trace downstream dashboards that will go stale. Fix with pool-slot recommendations.
- Five browser tabs at 2am
- Grep logs, ping the AE
- No downstream impact view
- One graph: AF → dbt → dashboards
- Agent reads logs in context
- MTTR under 2 hours
One map of the stack and ROI you can show
Living catalog with connector health. New hires productive in days. Canonical metrics end exec dashboard debates. Company Brain answers board prep.
- No single lineage view
- Weeks to onboard hires
- Metrics disputed in exec meetings
- Stack coverage at a glance
- Guided workspace onboarding
- One certified revenue number
Board-ready metrics without a data ticket
Query certified ARR, churn, and cohort KPIs via Company Brain. See lineage proof. Reconciliation reports when definitions change.
- Days waiting on analyst SQL
- Numbers change between decks
- No lineage for auditors
- Certified metrics on demand
- Cited lineage for every KPI
- Exportable audit trail
Dashboards that match the certified number
Publish canonical MRR from the metrics layer. BI tools pull certified definitions. Reconcile when finance flags a 2% delta.
- Three definitions of MRR
- No explore → warehouse lineage
- Weekly reconciliation tickets
- Certified metric in every tool
- Full BI → dbt → source chain
- One-click reconciliation
Feature stores with provenance you can defend
Trace features to source tables. Get alerted on schema drift. Know which training jobs consume each asset.
- Undocumented feature sources
- Silent schema changes
- Manual lineage notebooks
- Auto-discovered feature DAG
- Drift alerts from catalog
- Agent-assisted debugging
Self-serve product metrics without a SQL queue
Ask Company Brain in plain English. Get answers tied to product_event models. Trust certified activation and retention KPIs.
- Jira tickets to data team
- Stale numbers by ship date
- No trust in funnel metrics
- Plain-English metric queries
- Certified activation KPIs
- Drill into lineage if off
Policy and lineage without quarterly scrambles
Auto-discovered ownership and classification. Lineage exports for SOC2/GDPR. Impact analysis before policy changes.
- Static diagrams
- Manual PII tagging
- Quarterly audit scrambles
- Live ownership graph
- Exportable compliance lineage
- Audit log of sensitive queries
Regulatory lineage + fraud metrics you can defend
Self-host in VPC. Lineage from transaction models to regulatory reports. Certified fraud-rate and exposure KPIs.
- Manual examiner diagrams
- Fraud metrics in silos
- KPIs that don't reconcile
- VPC self-host · metadata-only
- Regulatory lineage exports
- Certified fraud & exposure metrics
PLG metrics the whole exec team trusts
Sync Segment + Snowflake + dbt. Certify NRR, logo churn, and activation. Company Brain for weekly business reviews.
- Churn defined 3 ways
- Product vs finance disputes
- Manual board slide prep
- Certified NRR & activation
- Lineage to board slides
- One number in every review
Revenue attribution across Shopify, ads, and warehouse
Connect Fivetran + dbt + Airflow in one graph. Debug inventory sync DAGs. Canonical GMV and contribution margin.
- Black Friday DAG surprises
- Marketing vs finance GMV
- Tab-hop to debug syncs
- Unified connector graph
- Certified GMV metric
- Pre-promo impact analysis
HIPAA-friendly workspace with clinical provenance
Deploy on-prem or private cloud. Tag PHI-adjacent tables. Lineage for trial cohort → outcomes models.
- PHI boundaries unclear
- Months-long audit prep
- Manual trial lineage docs
- On-prem deployment
- Auto PHI tagging
- Clinical pipeline graph
Watch time, ads, and rights on one catalog
Events, billing, and ads metadata in the semantic graph. Trace watch time to the title. Certified DAU, churn, and fill.
- Stream vs batch silos
- Engagement KPIs misaligned
- Blind to downstream impact
- Kafka metadata in graph
- Event → metric traceability
- Stream failure impact view
IoT ingestion and supply chain KPIs
Ingest IoT + ERP metadata. Lineage from sensor raw → OTIF dashboards. Canonical supply chain and defect-rate metrics.
- OTIF calculated 5 ways
- OT/IT data silos
- Blind schema migrations
- IoT + ERP in one catalog
- Sensor → OTIF lineage
- Pre-migration impact alerts
Built for your stack
40+ connectors. No rip-and-replace. Metroflow crawls metadata and builds understanding on top of what you already run.
Benefits you can put in a business case
What data teams see after connecting their stack in hours reclaimed, incidents avoided, and ROI.
Faster impact analysis
Half a day of Slack archaeology → 10-minute lineage query before every schema change.
Fewer escalations
On-call engineers resolve DAG failures with full stack context instead of tab-hopping.
Onboarding vs 3 weeks
New hires query a living catalog on day one not tribal knowledge docs.
Trusted revenue number
Canonical metrics end dashboard debates between finance, product, and BI.
Time to first insight
Connect sources, sync metadata, ask your first grounded question same session.
Data movement
Metadata-only crawls. Credentials and row data never leave your network.
Platform cost recovered in the first quarter through reduced incident time, fewer rollback PRs, and analyst hours reclaimed from ad-hoc lineage requests.
Open source at the core.
Deploy on your own terms.
Apache 2.0. Run on your infra. No vendor lock-in.
Frequently asked questions
No. Metroflow crawls metadata only: schemas, manifests, DAG definitions, and lineage. Your data never leaves your systems.
Generic copilots guess from prompts. Metroflow builds a semantic graph from your actual stack and grounds every answer in synced metadata.
Yes. The core is open source under Apache 2.0. Run via Docker in your VPC with full auditability.
Yes. The metrics and trust layer ships with the workspace. Define KPIs, standardize across dashboards, and query them through Company Brain from day one.
Connect your sources, wait for the initial crawl, and you're querying lineage and metrics in under 5 minutes for a typical dbt + warehouse stack.