Most teams are moving fast with AI agents right now. Experimenting, building, shipping. Governance is the thing they'll deal with later. Logan Kelly, CEO of Waxle AI, thinks "later" is arriving faster than most organizations realize.
In this episode, Logan joins Peter Maddison and Dave Sharrock to talk about what real agentic governance looks like, why bolt-on point solutions don't work, and how to give developers the freedom to build without letting things quietly go off the rails in production.
This isn't a conversation about slowing things down. It's about building a layer of visibility that the whole organization can use, not just the dev team.
What we cover:
Why one consistent governance layer beats a pile of separate tools
The difference between agent authentication, identity, and impersonation risk
What happens when multi-agent pipelines start compounding errors
How auto instrumentation reduces the burden on developers
Why PII and IP leakage are the governance gaps most teams miss
What "getting out of the experimentation phase" actually means before going to production
This week's takeaways:
Let your teams experiment. That's how you learn what agents can do. Just don't skip governance on the way to production.
Governance doesn't have to be a gate. The best version layers in without friction and gives everyone visibility, not just the dev team.
If a developer has to do extra work to implement a governance feature, that's a design problem. Good governance should work for the developer, not the other way around.
About the guests:
Logan Kelly is the CEO of Waxle AI, an agentic governance platform built to help organizations run AI agents safely in production environments.
Peter Maddison is Managing Partner at Xodiac Inc. and a certified ICF Coach with over 20 years of experience guiding business agility transformations, SDLC governance, and delivery optimization.
Dave Sharrock is the founder of IncrementOne and a Certified Enterprise Coach and Scrum Trainer with over 20 years of experience helping organizations work better. He has worked with Fortune 100 companies and early-stage startups across finance, manufacturing, education, and beyond.
About Definitely Maybe Agile:
Definitely Maybe Agile is a weekly podcast where Dave Sharrock and Peter Maddison cut through the noise around Agile, DevOps, and organizational transformation. No dogma. No buzzwords. Just honest, practical conversations from people who have done the work.
Agile and DevOps aren't the goal. They're tools to help solve real problems and deliver real value.
📩 Reach us at feedback@definitelymaybeagile.com🌐 Visit us at definitelymaybeagile.com
Timestamps:
00:00 Introduction and Welcome
00:44 Why AI Governance Suddenly Matters
01:22 One Governance Layer Beats Point Tools
04:28 Observe First, Then Add Policies
06:07 Agent Identity and Impersonation Risk
06:47 Auto Instrumentation Over Manual Checkpoints
09:24 Intent Versus Outcome for Control
10:11 Multi Agent Pipelines Multiply Failures
14:07 Common Triggers: Cost, PII, IP Leakage
18:14 Governance for Non-Developer Agent Builders
19:18 Retry Loops, Timeouts, and Server Hammering
23:54 What the Best Projects Look Like
26:17 Final Takeaways
#AIGovernance #AgenticAI #AIAgents #DevOps #DefinitelyMaybeAgile #EnterpriseAI #SoftwareDelivery #TechLeadership #Agile #DigitalTransformation