Platform · Overview
Block at the PR.
Quarantine at runtime.
Resolve with Zeus.
One contract declares what the data must be. The same contract gates the change, holds the rows that break it, and gives Zeus the context to draft the fix.
What’s in the platform
13 pages
The platform
Govern
Inside each area
12 capabilities
Contracts
Contract registry
YAML contracts, ownership, expectations and data-product context in one layer
Visual contract editor
Metadata, fields, quality rules and compliance posture without hand-writing YAML
PR-based quality gates
Contract changes reviewed in the delivery workflow, before they reach production
Observability
Quarantine explorer
Blocked rows, failed rules, affected products and recovery evidence in one view
Pipeline reliability
Failures, volume shifts, freshness drift and trusted-row impact across the estate
Rule analytics
Which rules fail most, which products are exposed, where remediation should start
Zeus work queue
Incident diagnosis
Alerts and ticket-driven requests become diagnosed items with context and a next action
Clear remediation paths
Routed explicitly as a contract PR or a direct remediation, never left ambiguous
Reusable playbooks
Repeated fixes promoted into operational memory, with confidence signals
Governance
Risk & compliance
Personal-data impact, quality coverage, classification and contract health
Audit evidence
A traceable record of changes, incidents, AI decisions, approvals and actions
Lineage with policy context
Upstream and downstream lineage tied to PII overlays, owners and rule violations
What changes
Operating model
| Moment | With LakeLogic | Without |
|---|---|---|
| A contract change | Reviewed as a diff, gated in CI | Merged, then discovered downstream |
| Bad rows at runtime | Quarantined at the boundary, counted and attributed | Loaded, then cleaned up later |
| Incident triage | Starts at blast radius and contract context | Starts at log archaeology |
| A fix | Drafted with evidence, approved by a steward · Design partner beta | Hand-written, then repeated next quarter |
| Compliance evidence | Exported from the contracts that enforce it | Assembled by hand before the audit |
| The visual contract editor | Authoring alongside the YAML · Beta | YAML by hand |
Runs on your stack
Metadata only
Engines: the contract executes here
- Apache Spark
- Polars
- DuckDB
Platforms: read and written in place
- Databricks
- Microsoft Fabric
- Snowflake
- BigQuery
- Redshift
- AWS EMR
- Glue
- Synapse
- Dataproc
Workflow
- GitHub
- dbt
- Airflow
- Jira
LakeLogic reads metadata: table and column names, schemas, lineage, row counts and run logs. Row values stay in your platform.
Run this on your stack.
See the operating layer against your own contracts and pipelines, or start with the Apache 2.0 engine and keep everything in your lakehouse.