Sign up to save your podcasts
Or
Share The Missing Layer: Why Your AI Agents Need a Governance Plane
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
The Missing Layer: Why Your AI Agents Need a Governance Plane
Your AI agents are running. But can you go to sleep at night knowing they won't send 3,000 emails to the same person, run up a $3,000 bill overnight, or give away refunds your business can't afford?
In this episode, Michael Fauscette sits down with Logan Kelly, CEO of Waxell AI, to explore why governance for agentic systems is not observability, not guardrails, and not something you bolt on after deployment. It's a dedicated plane that has to be there from the start.
TIMESTAMPS:
00:00 - Introduction
00:45 - Logan's journey from CallSine (autonomous sales agents) to Waxell AI
03:55 - What a governance plane actually does and why the distinction matters
07:06 - Why "we'll deal with governance later" is a dangerous bet
09:34 - 26 governance categories and 200-300 policies per agent fleet
10:20 - Deterministic observability vs. probabilistic agent governance
13:14 - Failure modes don't have to be terminal: the manager-employee model
14:05 - The Gong analogy: why agents need real-time coaching, not post-game review
16:42 - Making governance accessible to non-engineers
18:12 - Real-time semantic interception: filtering outputs before they reach the customer
21:17 - The air traffic controller analogy: governing upstream, not at the gate
23:11 - Business operations teams setting agent policies without writing code
25:43 - Digital employees: why business leaders need to be in the governance loop
28:00 - "Go to sleep at night" as the real value proposition of governance
29:47 - MCP layer security vulnerabilities and the rug pull attack
31:16 - Governance by design: building it into infrastructure from day one
34:07 - How governance actually makes agents faster and more creative
35:31 - Protecting customer experience and brand at machine speed
39:29 - The next 12-18 months: autonomy at scale vs. co-work at scale
43:02 - Thought leader recommendation: Dharmesh Shah (HubSpot)
ABOUT THE GUEST:
Logan Kelly is the CEO of Waxell AI, a governance plane for agentic systems. Previously, he founded CallSine, an AI-driven sales engagement platform that deployed fully autonomous agents, where he experienced firsthand the cost, quality, and security challenges that emerge when agents operate without a governance layer. That experience drove him to build the infrastructure that lets organizations run agents safely at scale.
CONNECT WITH LOGAN:
LinkedIn: https://www.linkedin.com/in/logankkelly/
Company: https://www.waxell.ai/
ABOUT DISAMBIGUATION:
AI clarity for business leaders. New episodes every Wednesday.
Host: Michael Fauscette, CEO & Chief Analyst at Arion Research
Author of "Building the Digital Workforce"
SUBSCRIBE & follow for weekly episodes on AI strategy, agentic AI, and enterprise technology.
Website: https://www.disambiguationpod.com/
Arion Research: https://www.arionresearch.com/
LinkedIn: https://www.linkedin.com/in/mfauscette/
#AI #ArtificialIntelligence #AgenticAI #AIGovernance #EnterpriseAI #AIStrategy #AIAgents #Podcast
In this episode, Michael Fauscette sits down with Logan Kelly, CEO of Waxell AI, to explore why governance for agentic systems is not observability, not guardrails, and not something you bolt on after deployment. It's a dedicated plane that has to be there from the start.
TIMESTAMPS:
00:00 - Introduction
00:45 - Logan's journey from CallSine (autonomous sales agents) to Waxell AI
03:55 - What a governance plane actually does and why the distinction matters
07:06 - Why "we'll deal with governance later" is a dangerous bet
09:34 - 26 governance categories and 200-300 policies per agent fleet
10:20 - Deterministic observability vs. probabilistic agent governance
13:14 - Failure modes don't have to be terminal: the manager-employee model
14:05 - The Gong analogy: why agents need real-time coaching, not post-game review
16:42 - Making governance accessible to non-engineers
18:12 - Real-time semantic interception: filtering outputs before they reach the customer
21:17 - The air traffic controller analogy: governing upstream, not at the gate
23:11 - Business operations teams setting agent policies without writing code
25:43 - Digital employees: why business leaders need to be in the governance loop
28:00 - "Go to sleep at night" as the real value proposition of governance
29:47 - MCP layer security vulnerabilities and the rug pull attack
31:16 - Governance by design: building it into infrastructure from day one
34:07 - How governance actually makes agents faster and more creative
35:31 - Protecting customer experience and brand at machine speed
39:29 - The next 12-18 months: autonomy at scale vs. co-work at scale
43:02 - Thought leader recommendation: Dharmesh Shah (HubSpot)
ABOUT THE GUEST:
Logan Kelly is the CEO of Waxell AI, a governance plane for agentic systems. Previously, he founded CallSine, an AI-driven sales engagement platform that deployed fully autonomous agents, where he experienced firsthand the cost, quality, and security challenges that emerge when agents operate without a governance layer. That experience drove him to build the infrastructure that lets organizations run agents safely at scale.
CONNECT WITH LOGAN:
LinkedIn: https://www.linkedin.com/in/logankkelly/
Company: https://www.waxell.ai/
ABOUT DISAMBIGUATION:
AI clarity for business leaders. New episodes every Wednesday.
Host: Michael Fauscette, CEO & Chief Analyst at Arion Research
Author of "Building the Digital Workforce"
SUBSCRIBE & follow for weekly episodes on AI strategy, agentic AI, and enterprise technology.
Website: https://www.disambiguationpod.com/
Arion Research: https://www.arionresearch.com/
LinkedIn: https://www.linkedin.com/in/mfauscette/
#AI #ArtificialIntelligence #AgenticAI #AIGovernance #EnterpriseAI #AIStrategy #AIAgents #Podcast
View all episodes
By Michael FauscetteYour AI agents are running. But can you go to sleep at night knowing they won't send 3,000 emails to the same person, run up a $3,000 bill overnight, or give away refunds your business can't afford?
In this episode, Michael Fauscette sits down with Logan Kelly, CEO of Waxell AI, to explore why governance for agentic systems is not observability, not guardrails, and not something you bolt on after deployment. It's a dedicated plane that has to be there from the start.
TIMESTAMPS:
00:00 - Introduction
00:45 - Logan's journey from CallSine (autonomous sales agents) to Waxell AI
03:55 - What a governance plane actually does and why the distinction matters
07:06 - Why "we'll deal with governance later" is a dangerous bet
09:34 - 26 governance categories and 200-300 policies per agent fleet
10:20 - Deterministic observability vs. probabilistic agent governance
13:14 - Failure modes don't have to be terminal: the manager-employee model
14:05 - The Gong analogy: why agents need real-time coaching, not post-game review
16:42 - Making governance accessible to non-engineers
18:12 - Real-time semantic interception: filtering outputs before they reach the customer
21:17 - The air traffic controller analogy: governing upstream, not at the gate
23:11 - Business operations teams setting agent policies without writing code
25:43 - Digital employees: why business leaders need to be in the governance loop
28:00 - "Go to sleep at night" as the real value proposition of governance
29:47 - MCP layer security vulnerabilities and the rug pull attack
31:16 - Governance by design: building it into infrastructure from day one
34:07 - How governance actually makes agents faster and more creative
35:31 - Protecting customer experience and brand at machine speed
39:29 - The next 12-18 months: autonomy at scale vs. co-work at scale
43:02 - Thought leader recommendation: Dharmesh Shah (HubSpot)
ABOUT THE GUEST:
Logan Kelly is the CEO of Waxell AI, a governance plane for agentic systems. Previously, he founded CallSine, an AI-driven sales engagement platform that deployed fully autonomous agents, where he experienced firsthand the cost, quality, and security challenges that emerge when agents operate without a governance layer. That experience drove him to build the infrastructure that lets organizations run agents safely at scale.
CONNECT WITH LOGAN:
LinkedIn: https://www.linkedin.com/in/logankkelly/
Company: https://www.waxell.ai/
ABOUT DISAMBIGUATION:
AI clarity for business leaders. New episodes every Wednesday.
Host: Michael Fauscette, CEO & Chief Analyst at Arion Research
Author of "Building the Digital Workforce"
SUBSCRIBE & follow for weekly episodes on AI strategy, agentic AI, and enterprise technology.
Website: https://www.disambiguationpod.com/
Arion Research: https://www.arionresearch.com/
LinkedIn: https://www.linkedin.com/in/mfauscette/
#AI #ArtificialIntelligence #AgenticAI #AIGovernance #EnterpriseAI #AIStrategy #AIAgents #Podcast
In this episode, Michael Fauscette sits down with Logan Kelly, CEO of Waxell AI, to explore why governance for agentic systems is not observability, not guardrails, and not something you bolt on after deployment. It's a dedicated plane that has to be there from the start.
TIMESTAMPS:
00:00 - Introduction
00:45 - Logan's journey from CallSine (autonomous sales agents) to Waxell AI
03:55 - What a governance plane actually does and why the distinction matters
07:06 - Why "we'll deal with governance later" is a dangerous bet
09:34 - 26 governance categories and 200-300 policies per agent fleet
10:20 - Deterministic observability vs. probabilistic agent governance
13:14 - Failure modes don't have to be terminal: the manager-employee model
14:05 - The Gong analogy: why agents need real-time coaching, not post-game review
16:42 - Making governance accessible to non-engineers
18:12 - Real-time semantic interception: filtering outputs before they reach the customer
21:17 - The air traffic controller analogy: governing upstream, not at the gate
23:11 - Business operations teams setting agent policies without writing code
25:43 - Digital employees: why business leaders need to be in the governance loop
28:00 - "Go to sleep at night" as the real value proposition of governance
29:47 - MCP layer security vulnerabilities and the rug pull attack
31:16 - Governance by design: building it into infrastructure from day one
34:07 - How governance actually makes agents faster and more creative
35:31 - Protecting customer experience and brand at machine speed
39:29 - The next 12-18 months: autonomy at scale vs. co-work at scale
43:02 - Thought leader recommendation: Dharmesh Shah (HubSpot)
ABOUT THE GUEST:
Logan Kelly is the CEO of Waxell AI, a governance plane for agentic systems. Previously, he founded CallSine, an AI-driven sales engagement platform that deployed fully autonomous agents, where he experienced firsthand the cost, quality, and security challenges that emerge when agents operate without a governance layer. That experience drove him to build the infrastructure that lets organizations run agents safely at scale.
CONNECT WITH LOGAN:
LinkedIn: https://www.linkedin.com/in/logankkelly/
Company: https://www.waxell.ai/
ABOUT DISAMBIGUATION:
AI clarity for business leaders. New episodes every Wednesday.
Host: Michael Fauscette, CEO & Chief Analyst at Arion Research
Author of "Building the Digital Workforce"
SUBSCRIBE & follow for weekly episodes on AI strategy, agentic AI, and enterprise technology.
Website: https://www.disambiguationpod.com/
Arion Research: https://www.arionresearch.com/
LinkedIn: https://www.linkedin.com/in/mfauscette/
#AI #ArtificialIntelligence #AgenticAI #AIGovernance #EnterpriseAI #AIStrategy #AIAgents #Podcast