Oct 8, 2026 · 32 min
Listen from 1:43
Listen at 1:43
AI agents need governance before companies scale them
EP 78: Do You Know What Agents You Have Running? Transcend's Rowan Stewart on AI Agent Governance
Unchecked agents can leave companies stuck in pilot purgatory while exposing them to compliance, customer, legal, and reputational risks.
- 1Companies often launch AI systems before establishing the controls needed to monitor, audit, and govern them.
- 2Deterministic permissions, observability, and agent inventories make increasingly opaque systems easier to manage at scale.
- 3Governance works best as infrastructure built early, reducing uncertainty instead of becoming a late compliance checkpoint.
Don't miss
Rowan Stewart recommends that companies first inventory every agent already in operation and map the systems and data each one can affect.
The brief
Rowan Stewart joins the hosts to examine why companies rush AI systems into production without the governance needed to monitor agents, control permissions, or trace their effects.
The result is often pilot purgatory: projects stall under compliance, customer, legal, or reputational risk because controls were treated as an afterthought.
As agents and language models become harder to audit, manually supervising every action becomes impractical, increasing the need for deterministic controls and observability.
Stewart’s central argument is that governance should function like infrastructure, with guardrails that reduce uncertainty and help companies move faster rather than merely slowing launches.
The episode’s clearest practical step comes at the end: inventory existing agents, identify the systems and data they affect, then govern before adding more.