The AI Agent 101 Episode: Agents, Skills & Shadow AI | Attempted Intelligence Ep. 1
What actually makes an AI tool an "Agent"—and why is "Shadow AI" quickly becoming every organization's biggest blind spot?
In the premiere episode of Attempted Intelligence, hosts Harlan Parrott (VP of AI Innovation at KnowBe4) and Matt Duren (SVP of Engineering for AI & Data) break down how practical AI has shifted from 18-month PhD projects to plain-language workflows every team can use. Built for practitioners, marketers, product managers, and executives who use AI at work every day, Harlan and Matt demystify the core terms driving modern enterprise AI: AI Agents, Skills, AI Slop, and Shadow AI.
Together, they explore the real definition of agency, how to safely enable AI tools without triggering Shadow AI risks, and why understanding these definitions matters now that regulatory frameworks like the EU AI Act are taking shape.
KEY TAKEAWAYS FROM EPISODE 1:
- The Evolution of AI: How LLMs unlocked agency and multi-step reasoning for everyday business workflows.
- Managing Shadow AI: Why "just blocking everything" fails and the 3-step framework for AI governance (Find, Enable, Block).
- Defining the AI Agent: Breaking down goal pursuit, variability, and giving up control over execution.
- AI Slop vs. AI Skills: How prompt engineering and skill refinement turn generic output into enterprise value.
CHAPTERS / TIMESTAMPS:
00:00 - Introduction to Attempted Intelligence
00:50 - Meet Your Hosts: Harlan Parrott & Matt Duren
01:11 - Two Perspectives: AI Across the Business vs. AI in Product Engineering
02:06 - An Introduction to AI for Practical Workers & Practitioners
05:11 - Rethinking Processes & Plugging AI In Judiciously
05:53 - AI Governance: Keeping Tools Safe While Delivering Value
08:29 - KnowBe4's AI Journey (Before ChatGPT)
11:37 - The Shift to Multi-Step Reasoning and Agency
17:44 - Why AI and Data Architecture Belong on the Same Team
19:20 - What Is Shadow AI? (And the Access Database Story)
25:14 - Why "Just Block Everything" Fails
26:15 - What Is AI Slop?
26:52 - How LLM Training Produces "Average" Writing
29:29 - Does "Built by AI" Automatically Mean AI Slop?
31:37 - What Is an AI Skill?
33:57 - What Is an AI Agent?
39:46 - Can End Users Tell the Difference Between Humans and AI Agents?
41:07 - Stitching Together a Working Definition of an AI Agent
41:58 - Why Definitions Matter: The EU AI Act & Governance
44:50 - Next Episode Preview: Data Centers & Society
FEATURED QUOTES:
💬 “When Microsoft Excel was first introduced... it was a new tool. It was so powerful people didn't know how to use it. And then over 20 years it transformed everything about how we work—and that's really what we're looking at now with AI.” — Harlan Parrott
💬 “Find where AI exists, enable it where you want it, and then block it where you don't. I think that order is really important.” — Matt Duren
RESOURCES & LINKS:
- Connect with Harlan Parrott on LinkedIn: https://www.linkedin.com/in/harlan-parrott/
- Connect with Matt Duren on LinkedIn: https://www.linkedin.com/in/matthew-duren/
- Discover how KnowBe4 secures the workforce across humans and AI agents: https://www.knowbe4.com/
- Subscribe to Attempted Intelligence for weekly episodes on applying AI to real work!
#AIAgents #ShadowAI #AIGovernance #EnterpriseAI #KnowBe4 #AttemptedIntelligence #ArtificialIntelligence #AIInBusiness