Agent Governance in Production: Control, Observability, and Human-Centered AI with Logan Kelly

Logan Kelly, CEO and co-founder of Waxell.ai, discusses building a governance, observability, and runtime control layer for AI agents in production so autonomous systems remain reliable, auditable, and aligned with business and ethical constraints.

Drawing on experience running autonomous sales agents at scale, he argues governance should sit above agents to avoid boilerplate, prevent bypasses, and let operations, finance, security, and frontline teams shape behavior without shipping new code. The conversation covers how governance can start with cost and quality controls, expand into runtime enforcement, and help organizations move faster through better visibility and safer adjustments.

They also address risks of employees bringing third-party agents into corporate environments, the need for organizational ownership and culture, and a future focused on augmenting humans with multiple agents rather than replacing workers.

00:00 Agent Governance Overview

00:53 Formative Lessons in AI

03:39 Why Governance Sits Above

05:50 What Customers Want First

07:07 Governance Helps You Move

10:19 Shadow Agents and Security

12:44 Who Owns the AI

15:26 AI Hype and Economics

19:32 Augmentation Not Replacement

25:42 Outcomes Over Process

28:58 Orchestration Over Copilots

30:02 Building The AI Foundation

31:05 Who Owns AI

33:12 Adoption And Culture

36:04 AI Needs Imagination

38:01 AI As A Skill

41:11 Guiding Teens For AI

48:53 Human Communication Edge

50:21 Five Year Agent Future

52:52 Cycling And Ambitions

https://waxell.ai/