Episode Overview
Podcast: AI Explored
Host: Michael Stelzner (Founder, Social Media Examiner)
Guest: Christopher S. Penn (Chief Data Scientist, Trust Insights; Author, Almost Timeless: The 48 Foundation Principles of Generative AI)
Episode Title: Setting the Stage for Agentic AI: A Practical Framework
Date: November 25, 2025
The episode dives deep into what agentic AI really means for marketers, creators, and business owners. It provides a practical framework for understanding and adopting agentic AI—AI systems capable of autonomous action—moving beyond hype to actionable strategy, tools, and next steps. The conversation clarifies misconceptions, explains implementation frameworks (done-by-you, done-with-you, done-for-you), and offers real-world examples and tool recommendations to accelerate AI integration.
Major Themes and Purpose
- Defining Agentic AI: Clearing up confusion and hype by giving marketers a grounded understanding of agentic (autonomous, action-taking) AI.
- Practical Implementation: Step-wise frameworks for getting started and scaling up agentic AI, from basic prompting to full autonomy.
- Actionable Tools & Tactics: Exploration of products (e.g., Opal, Claude Skills, N8N) and best practices for building agentic workflows.
- Future Outlook: Where agentic AI is headed and what marketers should be watching for.
Key Discussion Points & Insights
1. Misconceptions About Agentic AI (02:30–04:56)
- Widespread Confusion: The term “agentic AI” means different things to different people, fostering confusion and a marketplace ripe for overpromising vendors.
- Agent Explained With Analogies:
“A travel agent takes care of a lot of the minutiae so that you don’t have to worry about it... So when people talk about agents in AI...no one knows what you mean when you say AI agent or agentic AI.”
— Christopher S. Penn (03:20) - SEO Analogy: Current trends mirror past hype cycles—old solutions rebranded as “agentic” with inflated pricing.
2. True Definition & Value of Agentic AI (04:56–07:56)
- Agentic AI Explained:
Autonomous systems that take meaningful action for you—like “self-driving apps.” - Upside:
“It’s kind of like going from being a power user of AI to being a manager of AI, where now I’m managing a team of these agents that are off going, doing their thing.”
— Christopher S. Penn (06:40) - Scaling Impact: Delegates repetitive, complex, or multi-system operations to intelligent agents for scale and efficiency.
3. Three-Level Agentic AI Framework (08:26–11:32)
- Done By You: User does all the work. (e.g., prompting ChatGPT directly)
- Done With You: Hybrid—AI handles some tasks, user handles others (like meal kits or using custom GPTs/Claude Projects)
- Done For You: Full AI autonomy; system completes the process without user intervention (like dining at a restaurant).
4. “Done By You”: Foundational Practices and Prompts (11:32–18:00)
- Prompting Best Practices:
- One question at a time:
“Ask me one question at a time until you have enough information to successfully complete the task.”
— Christopher S. Penn (13:21) - Recap as system instructions:
“Recap the entire conversation as a set of system instructions for the next time using your prompt engineering knowledge.”
— C.S. Penn (14:20) - Ask for multiple options: Avoid single-answer tunnel vision, solicit multiple ideas/solutions.
- One question at a time:
- Framework Shared: The Casino Prompting Framework (Context, Audience, Scope, Intent, Narrator, Outline, Outcome).
5. “Done With You”: Mini-Apps, Workflows, and Emerging Tools (18:38–34:22)
- Mini-Apps & Projects:
Custom GPTs, Claude projects/skills, and Gemini “gems” serve as repeatable templates for repeated tasks. - Opal (Google):
- A “no code” workflow builder that links Google apps (Docs, Sheets, YouTube, Gmail, etc.).
- Live demo: Build an app to search news and make an explainer video, all with one prompt.
“If you’re a Google Workspace shop, that’s perfectly fine…there’s no infrastructure to host. It’s kind of a nice workflow system.”
— Christopher S. Penn (20:44) - Free & accessible at opal.google
- Claude Skills (Anthropic):
- “Skills” are like plugins. E.g., “Write like Mike Stelzner”—installable and shareable across teams.
- OpenAI Agent Builder & Microsoft Copilot Studio Flow:
- Both solutions provide workflow automation within their respective ecosystems.
- OpenAI’s option is described as “terrible” and immature (28:40).
- N8N (“N-eight-N”):
- Open-source automation platform (like Zapier/Make), recently enhanced with prompt-based workflow creation.
- Favored for privacy (run locally), growing in ease-of-use (cloud version now includes prompt-building).
- Podcast Use Case:
- Automated transcript pulling, quote identification, video clip extraction, and social media scheduling with almost no manual steps.
-
“[N8N] produces for me not just the summary… but also it does the work for me and just clips the video so that I don’t have to do the video editing. I can just take that and load it to my social media scheduler.” (33:54)
- Creative Ideas: Using AI for not just text, but identifying points of high energy in video, automating highlight clip creation.
6. “Done For You”: Full Delegation to Agents (36:08–43:12)
- Agent Systems Are Here:
E.g., Deep research agents in ChatGPT/Gemini, N8N agentic flows that “just do the thing and you never look at it again.”“Definition of done for you is literally that you don’t do anything other than pick up the results…someone shows up at your house with food, you’re like, oh, that was cool. You had to do nothing.”
— C.S. Penn (36:52) - Process Guidance:
- Don’t leap to full automation first—prove prompts and flows stepwise to minimize risk and catch flaws early
- Full delegation is only safe once confidence is established (like trusting a star team member)
-
“…you absolutely should not go straight to building an agent.” (37:05)
Tools & Real-World Example
- Claude Code (Anthropic):
- Build autonomous, multi-agent systems to, for example, author an entire book by aggregating and analyzing a content archive (e.g., export a subject matter expert’s LinkedIn, have four sub-agents build, edit, fact-check, and validate chapters).
“We hit go and walked away. And 92 minutes later, Claude had spit out a book, chapter by chapter, ready for her review.” (40:54)
- Now available via web; requires GitHub repository (not just for code—can use it for any document/collaborative project).
- Build autonomous, multi-agent systems to, for example, author an entire book by aggregating and analyzing a content archive (e.g., export a subject matter expert’s LinkedIn, have four sub-agents build, edit, fact-check, and validate chapters).
- Other Big Platforms:
- Google Vertex, Microsoft Azure, AWS—powerful, highly technical, developer-focused ecosystems.
7. Future Trends: Integration & Middleware (44:40–46:39)
- Ecosystem “Middleware” Is Coming:
Platforms and routers (like OpenRouter) are already mediating between models and tasks. - Software as Commodity:
The real differentiators for businesses will be people, processes, and unique data. - Rapid Evolution:
As ease-of-use matures, anyone will be able to build, direct, and deploy intricate agentic systems.
Notable Quotes & Memorable Moments
- On the Hype and Snake Oil:
“No one knows what you mean when you say AI agent or agentic AI. And so that to me is the biggest misconception of all… fertile grounds for snake oil salesmen to come in.”
— Christopher S. Penn (03:35) - Agentic AI in Context:
“Agentic AI means taking the engine, which is a model, and building the rest of the car around it…”
— Christopher S. Penn (05:16) - Prompting Wisdom:
“If you do nothing else, that will 2x your AI results immediately.”
— Christopher S. Penn (13:21, on "ask me one question at a time" prompt) - Autonomous Workflow Vision:
“It’s like going from being a power user of AI to being a manager of AI…”
— C.S. Penn (06:40) - Practical Automation Impact:
“Once the agentic portion…where it’s doing it without you, you know, because you’ve built it from the foundation up that everything…already works.” — C.S. Penn (37:05)
- On Full Autonomy:
“This is kind of the equivalent to having someone who works for you that you really trust and you just let them do their job and you kind of stay out of it.”
— Michael Stelzner (37:55) - Business Data as the New Moat:
“The defensive areas that you have as a business…are going to be around your people and your processes and your data, the data you have that other AI companies…don’t have…”
— C.S. Penn (46:19) - Closing Thought:
“We have just scratched the surface of that brain of yours…”
— Michael Stelzner (46:39)
Recommended Tools & Resources
- Opal (Google): opal.google — No code workflow for Google ecosystem
- Claude (Anthropic): claude.ai — Skills for mini-apps, Claude Code for agent ecosystems
- N8N: n8n.io — Workflow automation, local/cloud; supports prompt-building
- Trust Insights Casino Framework: trustinsights.ai/casino
- Prompt Recap & Conversation Systemization: Key for building repeatable agentic flows
- Exporting Social Posts/Data: All major social networks allow data export for source material
Timestamps for Important Segments
- [02:30] — Clearing up agentic AI misconceptions
- [04:56] — Defining agentic AI and real-world analogies
- [07:56] — Agentic AI as delegation and the product-market-fit analogy
- [11:32 – 18:00] — Foundational prompting practices for “done by you”; prompt framework details
- [18:38 – 26:56] — “Done with you”; Opal walkthrough, workflow design, and outputs
- [27:11 – 34:22] — Mini-apps (Claude skills, OpenAI, N8N), real-world automations, and creative possibilities
- [36:08 – 43:12] — “Done for you” is here; agentic flow design, real-world book writing agent
- [44:40 – 46:39] — Future: Middleware, integration across platforms, business data as moat
Final Takeaways
- Don’t be seduced by jargon—focus on agentic AI only when clear on your business needs and process maturity.
- Start with structured prompting, scaffold up to reusable mini-apps, then pursue full agentic autonomy cautiously.
- Current and emerging no-code/low-code tools make complex agentic workflows accessible, especially for small/solo teams.
- Your proprietary data and thoughtful process design are powerful differentiators in an AI-driven world.
- Continuous experimentation and literacy in prompt engineering are foundational for agentic AI success.
For more details or to connect with Christopher S. Penn:
- TrustInsights.ai
- YouTube, LinkedIn (@cspenn), and his blog ChristopherSpenn.com
Show notes and links: socialmediaexaminer.com/aipod
Episode transcript reference: socialmediaexaminer.com/A81
