# Database Governance for AI Agents

> Database governance applied to AI agents — identity, authorization, audit, and masking. The four controls that extend Bytebase's governance model from teams to ephemeral autonomous agents.

Source: https://www.bytebase.com/database-governance-ai-agents/

---

## Authenticated. Authorized. Audited. Masked.

AI agents access databases like a new kind of user — ephemeral, autonomous, at machine scale. The same governance you apply to teams, applied to agents.

## Four controls. From teams to agents.

### Identity

Each agent gets its own identity. Ephemeral, scoped, never shared with humans.

### Authorization

Just-in-time access. Granted per task, expired by default, never standing.

### Audit

Every query logged with the agent's intent and the human who initiated it.

### Masking

Sensitive columns redacted at query time. The agent sees what it needs, not the raw row.

## Read in order.

### Codify the context

What AI agents need beyond DDL. Schema as code is necessary, not sufficient. /blog/schema-as-code-to-schema-as-context

### Govern the access

Identity, authorization, audit, masking applied to ephemeral autonomous agents. /blog/how-to-govern-ai-agent-access-to-enterprise-data

### Apply to text-to-SQL

Context, evaluation, governance for enterprise text-to-SQL. /blog/enterprise-text-to-sql

## Get Started

- [Contact us](https://www.bytebase.com/contact-us/)
- [Start now (cloud)](https://console.bytebase.com)