AI Governance

Recommended reading order

1. AI Governance Roles and Ownership

   RACI for AI workflows using Northline B2B support-assist example—sponsor, process owner, IT, legal, and ops.

2. Data Boundaries for AI Agents

   Allow and deny matrices for agent tools—Northline B2B filled example for CRM, KB, email, and wiki.

3. Audit Trails for AI Workflows

   What to log for AI-assisted workflows—with sample JSON for Northline B2B ticket 4821.

4. AI Risk Review Cadence

   Standing AI risk forum with sample Northline B2B meeting minutes—cadence, agenda, attendees, and decision outputs.

5. When Prompt Injection Becomes an Action

   Untrusted content must not expand an agent’s authority. Check the execution layer, not only whether the model refuses.

6. Securing MCP and Agent Tools

   A production security playbook for MCP and agent tool access, including allowlists, permission scoping, and prompt...