Agent data governance
Who can query what, and can you prove it?
AI agent governance decides which data each agent may read and keeps a record of what it did. In Rig, Gateway gives each agent its own role, token and daily call limit. The agent reads certified metrics, sees only the schemas, columns and rows its role allows, and every call it makes lands in a named audit trail.
Most agents reach company data through a shared key
To give an agent data, teams paste a warehouse credential into a config file or connect it as an admin. The agent then sees everything, nobody can say what it queried, and its answers use whichever table it finds first.
What changes for each agent
Shared credentials
A role per agent
The role sets what the agent can read, the token identifies it, and the daily limit caps its calls.
Access to everything
Schema, column and row rules
Access control sets the schemas, columns and personal data an agent can read, and row rules limit it to the rows its role allows.
Queries nobody can see
A named audit trail
Rig logs every call an agent makes under the agent’s name.
Whichever table comes first
Certified metrics first
The agent looks up certified metrics and table descriptions, writes the query and tests it before it runs.
Changes nobody checks
Approval before changes
Changes to source systems, such as a CRM update, wait for a person to approve them.
Rules for each tool
One set of rules
The same permissions apply in Rig, Claude, Cursor and every other tool connected through Rig’s MCP access.
How the work runs
Agree the job
Agree what the agent does and which data it needs, with the business owner and the data team.
Set the role
Set the agent’s schemas, columns, rows and daily limit, and which changes need approval.
Give it context
Point the agent at certified metrics, table descriptions and your usage rules.
Review the trail
Read the audit trail, test results against the task, and adjust the role as the agent’s job changes.
Where to start
Common questions
What is AI agent governance?
It is deciding which data each AI agent may read and keeping a record of what it did. In Rig, Gateway gives each agent its own role, token and daily call limit, and logs every call under the agent’s name.
Who can query what, and how do you prove it?
Each agent’s role sets the schemas, columns and rows it can read, and owners can run SQL as the role to check before granting it. The audit log then shows every call with the agent’s name and tool.
Which agents does this cover?
It covers agents built in Rig, and agents in Claude, Claude Code, Cursor, Codex or ChatGPT that connect through Rig’s MCP access.
Can an agent change data in our systems?
Yes, but only through the actions you allow, and a person approves each change first.
How do we keep agent costs under control?
Each agent has a daily call limit, and the audit trail shows what it used.
Can two agents see different data?
Yes. Each agent has its own role, so a payments agent and a sales agent read different schemas, columns and rows.
What does the audit trail show?
It shows every call each agent makes, with the agent’s name, the query and the result, so the data team can review the work.
Talk to Rig
What work do you want to improve?
Tell us where your team spends time, loses money or needs better information.
In a 30-minute call, we’ll discuss the data and systems work that could help.