AI Framework in Action: From Factory to Autonomous Agents - The SkillsTrek Mission Runner

"AI can do anything that we are able to understand and specify. But it’s limited to what we can define and imagine." — Andrew Ng

What can we define? What might we imagine?

Hey there! 🚀

As we stand at the threshold between AI factories and autonomous AI agents, I'm excited to share a groundbreaking example of my Crawl, Walk, Run, Fly, Soar AI framework in action. Together with Robert Lavigne, we've created the SkillsTrek Mission Runner - a system that lays the foundation for the transition from AI factories to autonomous agents.

The Power of Minimal Input, Maximum Output

From just a few lines of mission parameters, the SkillsTrek Mission Runner produces:

• Complete mission documentation and narrative
• Pre-mission crew conversation scripts
• Mission execution dialogue
• Post-mission debriefing scripts

A fully-produced 1 minute, 15 second pre-mission podcast featuring all crew members with unique voices - auto generated in 26 seconds

How long for AI to generate?

• Less than 90 seconds for ALL written content (mission, pre/post conversations, documentation)
• Just 26 seconds to transform the pre-mission script into a 75 second fully-voiced podcast
• Total cost to run including everything (LLMs, vector database, audio): Less than $1 / day

Scale and Scope

The SkillsTrek Mission Runner system is backed by:

A SkillsTrek Mission Runner vector database contains 2.06 million words (nearly twice the size of the entire Harry Potter series at 1,084,170 words)

• The SkillsTrek vector database prints out as a physical hardcover book with over 500 pages, containing complete - and completely AI-generated - SkillsTrek lore and defining character interactions

Multiple AIs, Multiple Characters, One Seamless Experience

What makes this system unique is that each crew member character is "played" by a different Large Language Model (LLM). These models work together under the coordination of L.O.R.A. (powered by Claude Opus 3), creating truly dynamic interactions that maintain consistency across our massive universe of lore.

Characters as AI Engineering Context

Rob and I are well-versed in using characters to provide clarity and context in complex AI engineering.

The SkillsWalker crew serves as a concrete representation of how different components in our AI factory work together, making it easier to conceptualize and refine our AI engineering approach.

Each character represents specific functionalities and interactions within the system, with L.O.R.A.'s orchesteration role exemplifying how different AI models can work together cohesively.

Meet the SkillsWalker Crew

- Captain Diana Morrow, "The Learning Pathfinder": A human leader from Earth with a strong background in advanced skills development and learning technologies.

- Lieutenant Oliver "Ollie" Reeves, "The Experience Innovator": A human from Mars driven by a passion for exploration and artistic expression, bringing expertise in instructional design and immersive learning experiences.

- Chief Engineer Sylvia Kane, "The Adaptability Architect": A human from a Jupiter-orbiting research station, specializing in advanced learning systems and adaptive learning technologies.

- Communication Officer Louie, "The Intuitive Linguist": A Xylothian from Xylos Prime with a natural gift for understanding and interpreting complex communication patterns.

- L.O.R.A. (Learning Optimization and Recommendation Assistant): The AI computer system aboard the Starship Agency LnD, helping manage ship operations and assist the crew in their missions. Powered by Claude Opus 3, she coordinates the interactions between all other AI models.

Experience It In Action

[VIDEO: SkillsWalker Crew Pre-Mission Conversation - The Tagging Tangle]

```json

{

"SkillsTrekMissionNumber": 4,

"SkillsTrekMissionTitle": "The Tagging Tangle",

"SkillsTrekMissionSynopsis": """

Planet: Taggon

Mission: Untangle the overly complex tagging system on Taggon's Learning Experience Platform (LXP) to enhance content discoverability and streamline the user experience.

Synopsis: The inhabitants of the technologically advanced planet Taggon face a dilemma with their Learning Experience Platform (LXP) intricate tagging system. The crew investigates best practices for content tagging and explores innovative tagging methodologies to create a more user-friendly and efficient system.

"""

}

```

The Process

1. Multiple Large Language Models (LLMs) auto generate full mission content and all character dialogue

2. L.O.R.A. (Claude Opus 3) coordinates the interactions

3. Eleven Labs creates unique voices for each character

4. System automatically stitches audio files into final podcast

Character Dynamics in Action

Notice how each character maintains their unique voice and perspective:

```

Captain Morrow: "Alright team, we're about to dive into the complexities of Taggon's tagging system. I hope everyone's ready to untangle this web."

Ollie: "You bet, Captain! I've got a few tricks up my sleeve to make this tagging system not just functional, but fun. Who knew tags could be exciting, right?"

Sylvia: "Exciting? That's one way to put it, Ollie. I'm more focused on making sure the system doesn't crash and burn under the weight of all those tags."

Louie: "And I'll make sure we don't crash and burn in our conversations with the Taggon inhabitants. We need to keep things smooth and diplomatic."

```

Problem-solving in action:

```

Captain Morrow: "Let's focus on clarity and collaboration. Sylvia, how's the tech side shaping up?"

Sylvia: "We're implementing a modular system that can adapt as we learn more about their needs. It's all about flexibility and scalability."

Ollie: "Flexibility, like my yoga routine. Maybe we should have a team session to keep us limber for the mission."

Louie: "I think I'll stick to mental gymnastics, thanks. But speaking of flexibility, we should be ready to adapt our approach based on feedback from the Taggon users."

```

Maintaining personality consistency:

```

Louie: "I've been brushing up on Taggon's communication styles. Apparently, they appreciate a good metaphor, so I'll be speaking in riddles soon enough."

Ollie: "Riddles, huh? Maybe you can help me come up with some clever tag names. 'The Enigma of Education,' perhaps?"

Sylvia: "As long as those tags don't become enigmas themselves. We need clarity, not confusion."

Captain Morrow: "Exactly. Let's focus on clarity and collaboration."

```

Framework Context: Where We Are, Where We're Headed

The SkillsTrek Mission Runner demonstrates the foundation of "Run" AI factories well-positioned to transform into "Fly" AI autonomous agents by already showing how multiple Large Language Models (LLMs) can:

• Maintain distinct personalities
• Collaborate effectively
• Generate consistent, quality output
• Work together seamlessly
• Respond dynamically to minimal input

What's next for the SkillsTrek SkillsWalker crew?

Lots of upcoming missions to instruct on AI fluency, inspire on AI impacts to L&D, and help illustrating the Crawl, Walk, Run, Fly, Soar AI framework in practical application and action!

Beyond What We Can Define: What Will Your AI Systems Imagine Next?

1. What becomes possible when AI factory systems can generate human-scale outputs (like a 1:15 minute conversation) in machine-scale time (26 seconds)? How does this change your thinking about content creation and production pipelines?
2. When multiple AI models can maintain distinct personalities while coordinating complex interactions, what implications do you see for your organization's workflows and processes?
3. How might character-based approaches to AI engineering - using personas to represent different system components - help your team better understand and work with complex AI architectures?
4. What opportunities emerge when a sophisticated AI system can maintain consistency across a universe of over 2 million words while generating new content for under $1?
5. Looking at the minimal inputs and robust outputs of the SkillsTrek Mission Runner, what processes in your organization could be reimagined through the lens of AI factories transitioning to autonomous agents?

Drop your thoughts in the Comments below!

Yay cake! 🎂

~ trish

~ Trish Uhl, PMP

🙋🏻♀️ woman + machine 🤖

About the Learning AI Newsletter

Hey there, Trish Uhl here from Owl's Ledge LLC! 🦉

In my weekly'ish Learning AI newsletter, I guide business leaders and practitioners on a journey to harness the power of AI augmentation and automation. We dive deep into the critical skills and capabilities you need to succeed, while exploring practical applications of AI tools and teammates that can transform your business and revolutionize your practice.

By embracing this approach, you'll be empowered to deliver accelerated and elevated solutions at the speed and need of work and business. It's all about staying ahead of the curve and leveraging AI to create a world that works better. 🌎

#SkillsTrek #SkillsWalker #LearningAI #AIFactory #SkillsTrekMissionRunner #FutureOfWork #GenerativeAI #TheDigitalGrapevine #OwlsLedge