Metadata, not data movement
We crawl context and lineage. Row data and credentials stay in your network. Self-host is first-class, not a footnote.
Vision & Roadmap
Not another tool in the stack.
The intelligence layer above it.
Metroflow is the Company Brain for data teams: one governed graph of infrastructure and metrics so every agent, dashboard, and engineer starts from the same truth.
Now · Shipping
Next · Compounding
Later · Agentic
What we will not compromise
Public roadmaps work when the principles are visible, not buried behind clicks.
One brain for eng + business
Infrastructure graph and canonical metrics ship together. No “phase 2 metrics” product after a pure catalog.
Agents earn trust first
Answers with citations today. Governed action only after the graph, owners, and policies are real, not a chatbot with write access.
The path
Industry-standard Now · Next · Later: themes, not fake ship dates. Everything below is open on the page.
- Now Foundation
- Next Depth
- Later Execution
Public roadmap
Scan like a product board. No accordions. This page is the full story.
Now
Available at launch
Infrastructure + metrics intelligence in one open-source workspace.
Context-aware workspace
SQL, dbt, and Python editors grounded in your live stack, not a generic notebook.
Living semantic catalog
Schemas, ownership, and freshness that stay current as your stack changes.
Cross-system lineage
Warehouse → dbt → orchestration → BI on one graph. Blast radius before you merge.
Canonical metrics layer
Define KPIs once. Reconciliation and audit so finance and product stop dueling dashboards.
Company Brain
Technical + business AI with citations to real assets, not prompt-only guesses.
Self-host · Apache 2.0
Docker Compose today. Your VPC, your credentials, no data plane export required.
Next
Compounds with your graph
As Metroflow runs longer, trust and agents get deeper. Same product, more leverage.
Metric certification workflows
Finance sign-off paths, versioned definitions, and board-ready KPI packages.
Specialized agent library
Docs, tests, lineage, and ops agents with PR-ready outputs humans review first.
Governance integrations
Bring Collibra, Monte Carlo, Great Expectations-style signals into the same graph.
Business NL surfaces
Stakeholders ask KPI questions without SQL tickets, still bound by certified metrics.
Richer org onboarding
Team pods, ownership maps, and playbooks that turn tribal knowledge into queryable context.
Later
Agentic execution
From “what depends on this?” to “fix it safely,” on the intelligence you already built.
Agents as operators
First-class agents that propose changes with lineage diffs, tests, and owners attached.
Governed action environments
RBAC, policy, and audit on every automated step, with no black-box write access.
Safe apply paths
DAG fixes, model patches, and pool/retry recommendations with human-in-the-loop gates.
Continuous stack health
Agents watch drift and incidents with the same graph that powers Q&A today.
The flywheel
Intelligence compounds.
You never start over.
Every crawl, certification, and ownership tag deepens the Company Brain. Launch features use that graph today. Next-horizon agents act on the same foundation. One platform, continuous context.
- Connect: warehouse, dbt, orchestration, BI
- Understand: lineage, metrics, ownership
- Trust: certified KPIs, citations, audit
- Act: governed agents on proven context
What this means for your team
You can plan without vaporware. Everything under Now is what you can install and run. Later themes are direction, not promises on a calendar that slip.
Your feedback moves the board. Priorities inside Next and Later shift with real teams and open-source input, while the foundation you buy into today stays complete.
You won’t re-platform for agents. Execution builds on the same Company Brain you self-host now. Context, metrics, and trust carry forward, with no second product to rip in later.