RAI(2) EP9: Guardian AI, hybrid AI agents, & brand-safe automation, w/Brian Ritchie, CEO & Founder

In this episode of Responsible AI in Action, Brian Ritchie breaks down the emerging concept of Guardian AI and how it connects to kama.ai’s Sober Second Mind approach. He explains how Guardian AI can act as an oversight layer for other AIs, especially as the industry moves toward more autonomous, agentic AI and approaches artificial general intelligence.

Brian describes kama.ai’s hybrid AI agent design, where a deterministic knowledge-graph “left brain” only delivers curated, authorized enterprise knowledge, while a generative “right brain” uses retrieval augmented generation from trusted collections. He also introduces Emotion Graph RAG, a patent-pending capability that uses human values as a behavioural model to evaluate both user inquiries and AI responses before they are delivered.

Key themes in this episode:

• What Guardian AI is and why Gartner and others are paying attention
• The difference between hybrid AI agents and fully agentic AI
• How a human-value–guided knowledge graph can act as a human behavioural model
• Emotion Graph RAG and applying human values to AI inputs and outputs
• Preventing harmful or high-risk responses (for example, 3D-printed weapons) before they reach users
• Governance, auditability, and “risk sliders” that let enterprises decide where generative AI can safely be exposed
• How responsible AI and Guardian AI can become a trust and brand differentiator

Whether you are building AI agents, setting governance policies, or exploring how to scale AI responsibly, this episode offers a clear, practical view of how Guardian AI can support safer, more transparent deployment.

Watch the full episode to learn how hybrid AI and Guardian AI can work together to keep AI aligned with human values.

Connect with kama.ai and Brian Ritchie:
Website: kama.ai