Harnessing Collective Intelligence with Digital Twins

“What these tools allow you to do is very, very quickly go from an idea to sort of an 80% manifestation of it. It’s not just about the technology—it’s about understanding how, when, and why to use it to unlock collective intelligence.”

– Kyle Shannon

“We’ve discovered you can externalize the voice in your head into something you can have a dialogue with, creating reflective moments that result in documentation, not fleeting thoughts. That’s transformative.”

– Kevin Clark

What you will learn

• Exploring the power of digital twins in collaboration
• Overcoming creative blocks with generative AI tools
• Asking better questions to unlock AI’s potential
• Designing structured interviews for personalized AI
• Understanding collective intelligence in the digital age
• Rapid prototyping to test and refine ideas quickly
• Reshaping industries with untapped organizational data

Episode Resources

• Emily Shaw
• Aristotle
• Steve Jobs
• Content Evolution
• CoLab
• Storyvine
• AI Salon
• Fortune 500
• Gartner
• Digital twins
• Generative AI
• Large Language Models (LLMs)
• GPT
• Notebook LM
• Transformer architecture
• Data collaboratives

Books, Shows, and Titles

• Collective Intelligence and AI 
• Candy Ears
• The Hitchhiker’s Guide to the Galaxy

Transcript

Ross Dawson: Ross, Kevin, and Kyle, wonderful to have you on the show.

Kevin Clark: Pleasure to be here.

Kyle Shannon: Ross, great to be here.

Ross: So, you created a book recently called Collective Intelligence and AI. I’d like to pull back to the big picture of where this fits into what you’re doing. This organization is called Content Evolution. How did you get to this place of creating this book and the other things you are doing using AI to assist in your work?

Kevin: Well, Content Evolution itself is a federation of companies that are aligned. We’re all thoughtful leaders and innovators and have been at it for 23 years now. This technology is helping us pull the thread forward a lot faster.

As Kyle will describe in a moment, we have almost 30 digital agents—or what we call digital advisors—of ourselves. As a result, we have a collective of those, and we can all write together. We’ve published articles and done all kinds of things. This book is a particular expression between the two of us because we’ve been talking to each other for over a decade. It’s the residue of a decade’s worth of weekly conversations. There’s more to it—Kyle, say more.

Kyle: When we started, we put together a group within Content Evolution called CoLab. The initial idea was, “Hey, this AI stuff is happening.” We started this probably a year and a half ago, almost two years ago. Generative AI was clearly evolving rapidly, so it felt important to explore.

Like with all new technologies, you start with the tools, but very quickly, you ask, “Why? What are we trying to accomplish?” Content Evolution is an organization that’s a couple of decades old. One challenge was figuring out who’s in it and what talents exist within it.

Initially, we asked, “Could we create a tool using generative AI to help someone discover the right person for a business problem?” That’s how it started. Over time, we realized we could create digital representations of ourselves—digital twins or digital advisors—that people could interact with 24/7. Even if Kevin wasn’t available, you could get his point of view.

We’ve built 30 of these digital twins. They’re all in a single entity, a single GPT, where we can query them for the Content Evolution perspective on a topic. Individuals within that group can also comment on outputs. A big part of what we’re exploring now is understanding how, when, and why to use these tools. That’s far more fascinating than just the technology itself.

Kevin: By the way, Kyle is the world’s first Chief Generative Officer. We didn’t put AI in the title because being generative is more important than the specific technologies you use. It’s about the practices, methodologies, and discernment of when to apply them—and sometimes, when to set them aside.

We’ve discovered you can overcome writer’s block quickly by having a prompted start for something you’re thinking about. We’re also learning to externalize the voice in our heads into something we can have a dialogue with. This creates reflective moments and produces documentation rather than fleeting thoughts. Fascinating, isn’t it?

Ross: Absolutely. The title Chief Generative Officer feels more appropriate, given the context. AI is just a set of tools.

Kyle: Exactly. You can generate content with the tools or on your own. It could even be a hybrid. You can also generate revenue or other outcomes. The generative aspect goes beyond just the tools.

Ross: The questions you raised are exactly the kinds of questions I wanted to ask. Starting with the basics, how are these digital twins set up? Are they based on system prompts or custom instructions for commercial LLMs?

Kyle: Right now, they’re custom GPTs, but we’ve experimented with other platforms like Poe and Claude. Initially, we wanted to scrape LinkedIn profiles to discover expertise within Content Evolution. But we realized a LinkedIn profile is a very thin, historical slice of who someone is. It doesn’t reflect how they talk, think, or solve problems.

We designed a structured interview with 27 questions across various categories. This interview digs into who someone is today, their inspirations, problem-solving approaches, worldview, and more. The answers to these questions form the foundational data for a custom GPT with a tailored prompt.

Ross: So, for someone in your network, do you conduct a voice or text interview for these questions?

Kevin: That’s a great question because there’s a difference.

Kyle: We learned that when people wrote their responses in text, their digital twins turned out horrible—just bad. People don’t write the same way they talk.

We now conduct video interviews where we go through the structured questions interactively. As the interviewer, if I notice someone hasn’t gone deep enough or gets excited about something but cuts themselves off, I’ll ask them to expand. Once we made this interactive, the digital twins came to life.

Kevin: It takes about 45 minutes to complete the interview. The questions are designed to be unusual, going beyond superficial answers. People are often surprised by the depth of the questions.

Kyle: One of my favorites, which was developed by Joke Gamble, is: “Describe your career in three acts.” It frames the career as a journey or drama, putting you in a different mental space. The quality of the questions is everything.

Kevin: Exactly. The quality of the question determines the quality of the answer from a large language model. At Content Evolution, our original tagline was “Be Intentional.” For 20 years, we’ve challenged our clients to ask better questions. That’s what we’ve been practicing all along.

Kyle: Asking better questions is the core of being a good prompt engineer. It’s about having expertise but also being able to communicate across disciplines. Our team members have this cross-disciplinary ability, which makes us well-suited to leverage this technology.

Ross: That’s a key point. Even though the answers from LLMs are improving, the most important thing remains the question. It reminds me of The Hitchhiker’s Guide to the Galaxy—you may know the answer, but asking the right question is crucial.

Kyle: Exactly. Inside the Heart of Gold with the improbability engine, you never know what’ll come up.

Kevin: Right. I’d also argue that this technology is redeeming the liberal arts degree. It enables specialization across disciplines, encouraging lifelong learners to embrace a generalist perspective. It’s about knowing how to organize and synthesize human knowledge.

Ross: Absolutely. Humans excel at synthesis, and now we have access to diverse ideas that nurture that capability. From the structured interviews, how do you translate the data into a GPT?

Kyle: We made strategic decisions for our official Content Evolution digital advisors. All of them share the same structural data: the interview forms the core, and every twin has the same system prompt. If we update the core prompt, it applies to all of them. The collection of 30 twins also has its own prompt.

Some members have created duplicates of their twins and added their writings, articles, books, and papers. These are different types of GPTs—one captures the person’s essence, and the other their body of work. It’s fascinating because the core data makes the twins inherit the personalities of the people behind them.

Kevin: Here’s a fun example. Kyle met a podcaster, Emily Shaw, who has a show called Candy Ears. She experimented with our digital twins, taking voice samples to mimic how we sound. Then she asked our twins questions and recorded their answers.

Kyle: We first answered the questions ourselves. Then she played the twins’ responses, and we rated them.

Kevin: I rated my twin a 7.5 out of 10. My wife, Heidi, said it sounded just like me and thought it deserved a 9 or 10. She’s lived with me for almost 50 years, so I’ll take her word for it! The question was something broad, like, “What is the meaning of life?” The alignment between my response and my twin’s was striking.

Kyle: For me, the text responses were spot on. However, the voice delivery didn’t match my dynamic range—I talk loudly, softly, quickly, and slowly. For someone with a monotone style, the twins are nearly identical.

Ross: Voice rendition is a challenge, but we’re on the verge of improving it. Kevin, you mentioned earlier that you use this group of 30 digital twins collectively. How does that work?

Kevin: All the individual twins are in a common folder labeled “CE GPT Profile Complete.” When I write an article for LinkedIn, I can query the folder: “Who in the community would have something to say about this?” It pulls relevant quotes and drafts an article, complete with an executive summary and attributions like, “Kyle says this,” or “Cindy Coon says that.”

Before publishing, I share the draft for feedback to ensure accuracy. Even if people don’t actively use this technology, engaging with it leaves a residue that makes them better. For instance, I couldn’t spell well growing up, but using spell check gave me immediate feedback and improved my skills. Similarly, interacting with this tech enhances capabilities over time.

Ross: So these are custom GPTs fine-tuned with your methodology?

Kyle: Yes, that’s correct. They’re private but also available in the GPT store for public interaction as part of our marketing. People can experience what Kyle Shannon or the collective might say on various topics.

Kevin: We also host a weekly program called Content Evolution: New World, where people can call in. Sometimes, we feed the transcripts into the GPT profile to generate LinkedIn posts summarizing the discussion. It does a decent job turning an hour-long conversation into a seven-paragraph post.

Ross: Kyle, you mentioned the book Collective Intelligence and AI. What’s the process from idea to a finished, shippable product?

Kyle: Kevin often says the book reflects a decade of our conversations. We meet weekly, and I’m the CEO of Storyvine, where Kevin is our senior advisor. This collaboration has been ongoing for years.

Personally, when I get excited about new technology, I dive in. Large language models initially felt counterintuitive—simple probability calculators, yet producing outputs that felt human. One day, I saw a tweet: “Artificial intelligence is the collective intelligence of humanity.” That hit me. The magic isn’t in how the tool works; it’s in what it’s trained on. I realized it allows us to collaborate with everyone who’s contributed to the internet.

I shared this insight with Kevin, and it sparked deeper discussions about collective intelligence—not just in machines but also in our CoLab. The idea evolved, and tools helped us quickly go from concept to an 80% draft.

Kevin: After that conversation, I went into the tools, wrote some prompts, and told Kyle, “I just outlined this as a book. What do you think?” He mentioned a tool that could write the whole thing, but I wasn’t interested in going that route. I’m more of a policy person, while Kyle dives into current trends. He also has his community, the AI Salon, which is very popular with lots of opt-ins.

We fed our manuscript into Notebook LM. It provided an interesting summary, but it also generated profound insights we hadn’t written. One example was: “The authors are saying it’s like being given access only to the children’s section of the library, without reading the adult books.” That was exactly the point. Much of human knowledge—especially advanced knowledge—is inaccessible because it’s behind firewalls, paywalls, or hasn’t been digitized. We’re only at the beginning of this journey.

Ross: That’s such a compelling metaphor—children’s versus adult sections. There’s so much knowledge that remains untapped because it hasn’t been captured or digitized. It’s an important insight.

Kyle: Agreed. One of the things we’ve written about is data collaboratives. Creating shared data lakes is crucial for organizations to think about and act on.

Ross: What are some examples of data collaboratives you’ve seen or worked with?

Kyle: The concept isn’t new—trade associations are a simple example. They bring together organizations with common interests, enabling them to share best practices without crossing legal boundaries. Large consulting firms also facilitate sharing across industries while respecting confidentiality.

AI accelerates this process because it doesn’t care about your industry—it can recognize parallels, analogize, and bring insights to bear faster than ever before. It just needs a prompt to get started.

Kevin: What amazes me about AI, particularly transformer architecture, is how it can hoover up enormous amounts of data and derive value with enough compute power. My organization has been around for over a decade. If I think about all the knowledge trapped in PowerPoint presentations, sales documents, and more, it’s substantial. We could plop all of it into an AI model and instantly gain insights.

Now imagine a Fortune 500 company or a trade association pooling their data. The value trapped in unstructured formats is immense. With just a little organization, they could unlock incredible potential.

Kyle: Often, this data sits on individual hard drives, disconnected from the cloud. Gartner predicts that in the next five to seven years, employment agreements will include clauses allowing companies to replicate your work processes and contributions. This will become part of the terms and conditions for employment.

Ross: That’s a fascinating point. To wrap up, what’s the generative roadmap for Content Evolution? What’s next for Kevin and Kyle?

Kyle: One thing I’m excited about is using the collection of digital twins to explore ideas in unique ways. For instance, if we have a new piece of legislation or an article, we can query the twins for 10 different perspectives—some close to my thinking, others wildly different.

We’re now working on a system that allows us to collaborate with people based on how they think and solve problems, rather than just their professional expertise. I can have a brainstorming session with people similar to me or choose those who think completely differently to challenge my ideas. This could even extend to historical figures—where would Aristotle or Steve Jobs sit on that spectrum? That’s what excites me.

Kevin: Let me add to that. On Tuesday, Kyle and I had a conversation that ended at 10:55 AM. By noon, Kyle had already prototyped and demoed the idea we discussed. That’s the power of rapid prototyping—there are no bad ideas because you can quickly test them.

Another key aspect is transcending limitations like time zones or language barriers. Right now, you can’t always get on someone’s calendar. But with digital twins, people can access our knowledge anytime, in their preferred language, and then decide if they need to speak to us directly. This approach transforms business and how we engage with the world.

Our challenge is often being so far ahead of the curve that people initially don’t understand what we’re talking about. That’s part of the innovator’s dilemma. But we’re excited to keep pushing forward.

Ross: That’s fantastic. We’ll include links to everything in the show notes. Where can people learn more about what you’re doing?

Kyle: Visit contentevolution.net. One of the first tools we built there is the Challenge Engine. You input a business challenge, and instead of giving answers, it generates questions to guide your thinking. You can also find us on the GPT Store by searching for “CE Profiles.”

Kevin: For those interested in staying updated on this space, I highly recommend Kyle’s AI Salon. It’s a vibrant community discussing AI and its implications. Kyle, where can people find it?

Kyle: The URL is thesalon.ai. We host bi-monthly meetings featuring speakers and discussions. The focus is on exploring what we can do with AI now that it’s accessible to everyone—not just engineers and mathematicians.

Ross: Great. Thank you so much for your time and insights, Kevin and Kyle. It’s been wonderful hearing about your work.

Kyle: Thank you, Ross. It’s been great to be here.

Kevin: Absolutely. Thanks, Ross.