Sempleo Docs
Guides

Governance

Configure approval gates, tool restrictions, and audit logging for enterprise AI governance.

Governance features are available on the Professional and Enterprise plans.

Sempleo provides enterprise governance controls to ensure AI agents operate within your organization's policies.

Approval Gates

Approval gates require human review before an agent can take certain actions:

How It Works

  1. An agent determines it needs to perform a governed action (e.g., sending an email, creating a Jira ticket)
  2. The action is paused and a notification is sent to the designated approver(s)
  3. The approver reviews the proposed action and approves or rejects it
  4. If approved, the agent completes the action; if rejected, the agent is notified

Configuring Approval Gates

  1. Go to Settings → Governance
  2. Enable governance for your workspace
  3. Define which actions require approval:
    • Integration actions (send Slack message, create Jira issue, etc.)
    • External API calls
    • Knowledge base modifications
  4. Set approvers (specific users or roles)

Tool Restrictions

Control which tools agents can use:

  • Allowlist — Only specified tools are available
  • Blocklist — All tools except specified ones are available
  • Per-team — Different tool access for different teams

Audit Logging

Every agent action is logged with:

  • Timestamp — When the action occurred
  • Agent — Which agent performed the action
  • User — Who triggered the agent
  • Action — What the agent did
  • Input/Output — The data involved (redacted for sensitive fields)
  • Approval status — Whether approval was required and granted

Viewing Audit Logs

  1. Go to Settings → Audit Log
  2. Filter by date range, agent, user, or action type
  3. Export logs as CSV for compliance reporting

Retention

PlanRetention
Trial7 days
Starter30 days
Professional90 days
Enterprise365 days (configurable)

Model Governance

On the Enterprise plan, you can:

  • BYO LLM — Use your own Azure OpenAI or AWS Bedrock deployment
  • Model restrictions — Control which models agents can use
  • Data residency — Ensure data stays in your preferred region

On this page