One decision layer for every data platform
Policies, meaning, and access rules as versioned operational metadata. Connect your databases and warehouses — one governed catalog, one set of decisions.
How It Works
Connect a platform, author decision-grade policies with AI, deploy them as live decisions, and audit the workspace lifecycle.
Connect Your Data Platforms
Point Metatate at PostgreSQL — including Supabase, Neon, Amazon RDS / Aurora, Google AlloyDB, and Timescale — MySQL / MariaDB, Google BigQuery, Amazon Redshift, Databricks, or Snowflake. Metadata sync and connection tests run worker-side; credentials go straight to the vault.
Author Policies With AI
Describe a policy in plain language and the AI assistant compiles it into decision-grade instructions — classification, masking, role grants — each marked enforceable or advisory. Draft-readiness checks and a completeness score tell you what's missing before anything ships.
Deploy Live Decisions
Approve a version and publish it. Every deployment shows what is live, where it applies, and why it is live — the source policy, instruction, and conflict status behind every serving decision that MCP answers rely on.
Audit Every Event
The activity log captures the full lifecycle — policies created, submitted, approved, and archived, members invited, roles changed, data sources added — filterable by event type, with the actor and details on every entry.
Connect an AI agent in one command
Every workspace exposes a Metatate MCP endpoint. Point Claude Code, Cursor, or any MCP-capable client at it with a workspace-scoped bearer token, and your agents get the governance tools — discover context, authorize use, validate queries.
- One endpoint serves every workspace — your token scopes it
- Governance tools registered out of the box — reads by default, review requests only via an explicitly scoped token
- Works with Claude Code, Cursor, and any MCP-aware client
Operational metadata as a decision layer
Structured metadata that agents can parse, query, and act on — consistent across every platform you connect.
One Governed Catalog
Tables from every connected platform land in a single catalog with meaning, classification, and sensitivity — one place agents and people look before they act.
AI-Assisted Authoring
Plain language in, decision-grade instructions out — classification, masking, and access rules, each marked enforceable or advisory, with completeness scoring and a draft-to-approval workflow.
Agent-Ready MCP Server
Agents query and reason over structured context through the Model Context Protocol — the same governance tools across every platform you connect.
Governance Without Bottlenecks
Policies deploy as live decisions, each marked advisory or enforceable. MCP answers stay consistent because they read the same layer.
Where Metatate meets your tools
Your decision layer has two sides. Agents and dev tools read it through MCP; the data platforms underneath are governed by it. Both plug into the same layer.
Claude Code plugin
Slash commands and a Claude skill that keep Claude grounded in your decision layer. Works against any Metatate deployment through MCP.
Cortex Code plugin
Same slash commands inside Snowflake's native developer workspace. Talks to the Snowflake-managed MCP — your data and audit never leave the account.
Any MCP agent
Metatate speaks the Model Context Protocol. If your agent framework is MCP-aware, it already reads the decision layer — no plugin required.
Nothing to Run
Fully managed. Connect a data source and go from zero to governed in minutes — start free, no commitment.
Read-Only by Design
Metatate syncs metadata, not data. Credentials are vaulted, connections are tested worker-side, and tokens are scoped per workspace.
Versioned & Portable
Policies are versioned operational metadata. Evolve them over time and serve the same decisions to every platform and agent.
Every platform.
One set of decisions.
Start free with unlimited policies, agents, and MCP requests — pricing grows with your infrastructure, not your activity.
Start free