How to build a digital twin with ChatGPT

Create (and interact with) a digital twin to get a better understanding of your customer, your character, or even yourself.

There’s no bigger barrier to getting something shipped than a fear over how our ideas and outputs will hit. Whether an offer receives crickets or fanfare, a pitch gets shot down or eaten up, or someone is going to take feedback as a criticism or an attack. It’s a fear that’s kept me (and most of us) from sharing or saving a lot.

One of the best ways through this is getting feedback. In other words, getting a sneak peek into what someone might say (or do) when asked a given question or in a specific circumstance. The response to which builds our confidence and adds a few degrees to the never-complete 360-degree understanding of a person or segment.

---

Key Takeaways

• Digital twins provide a fast, low-cost way to simulate feedback and reactions from personas, people or characters

• Building a digital twin requires curating diverse data about behavior, preferences, and/or communication style

• Approaches to using digital twins include simulation testing, objection handling, A/B testing, and narrative development

---

Enter digital twins, as they’re called, or the use of AI as a proxy for bouncing around ideas, themes, and conversations. They’re especially helpful to marketers, novelists, and anyone working within (or leading) a team, to name a few.

In this post, we’re going to talk about:

• Why digital twins offer speed and confidence

• How to create one with ChatGPT (including the data and prompts you’ll need)

• Patterns for interacting with them in the world of writing, marketing, and the workplace

The cost of getting feedback

Feedback is everything. It can make or break an initiative, accelerate a career, or be the difference between the trash heap or publication. But the nuance is in the how. The difficulty of getting feedback in traditional ways (from humans) is three-fold:

1. We spend (if not waste) valuable social capital: When we ask for feedback, we’re asking someone to give us their time and mental capacity. That’s a bank that we can only pull from so many times before the well dries.

2. We waste already thin attention spans: Giving feedback requires someone’s attention. The quality of what we get back is a factor of the attention they have to give.

3. Feedback cycles take a long time: Add both of the previous things together, and delivering feedback (good, bad, or any) is typically at the end of someone’s to-do list. The result is weeks to months to get any sort of feedback.

That’s where a digital twin benefits us, we can get feedback (especially early) that strengthens our approach to the task at hand. Does it mean we don’t need feedback from humans? Likely not, but in some cases, sure. But early preparation run through a digital twin can indeed make feedback cycles with humans more effective.

And while the digital twins don’t save us from the all-important research of knowing who we’re targeting (or creating, in the case of story telling), they allow us to cycle farther and faster once we do.

---

Where we can use digital twins

Digital twins simulate thinking (and if needed, tone). Here are three high-impact areas where digital twins can accelerate our work:

1. Marketing and sales

We can simulate how different personas might respond to products or messages. This unlocks smarter, faster decision-making in our strategies:

• Pre-sell testing: Your twin can generate synthetic feedback on messaging, helping you refine tone, emotional appeal, or clarity.

• Persona modeling: Simulate how power users vs. skeptics will respond to launch emails, landing pages, or ads.

• Content generation: Draft posts, lead magnets, and launch assets in your brand voice, vetted through the lens of your ideal customer.

2. Novel writing and storytelling

For authors, a digital twin can be both a test audience and a tool for immersive character-building.

• Character development: Simulate a character’s internal logic and tone, to keep actions and dialogue believable and consistent across their entire story arc.

• Reader response forecasting: Predict how a reader might react emotionally to a twist or chapter ending, helping fine-tune pacing and hooks.

• Query alignment: Validate whether a manuscript query or a story pitch will resonate or fall flat.

3. Workplace productivity and culture

Digital twins can help knowledge workers simulate interpersonal dynamics and communication patterns.

• Feedback personalization: Test how a teammate or direct report might receive constructive feedback based on their profile or past behavior.

• Meeting and messaging optimization: Simulate how different framing of messages affects clarity, motivation, or buy-in.

• Change management strategy: Try out different communication plans for organizational shifts to gauge barriers or confusion.

As a safe, intelligent space to simulate reactions, digital twins let us fine-tune before it ever hits human ears. Whether you’re shipping a product, a novel, or a message, it’s low leverage and high insight.

---

Building your digital twin in ChatGPT

Let’s start with “Why ChatGPT”? Like I’ve previously discussed around creating your own marketing assistant, ChatGPT works better than a tool like Claude for its ability to handle large threads (as well as very long individual prompts). The idea is that I’ll keep coming back to this twin over weeks, months, or years, so that’s a must-have.

Now, on to creating our digital twin. It isn’t about replicating perfectly—it’s about building a proxy that understands the perspective and decision-making of the actual persona or person. Once set up, this twin becomes a voice on the other side of the table, capable of pressure-testing ideas, role-playing, and obtaining faster feedback.

Here’s how to build your own.

Step one: Gather the data sources for your digital twin

Before you “build” anything, you need the raw material—patterns, preferences, and POV. Collect content that reflects the thinking and/or speaking of your person/persona. This can come from any number of sources, and can include:

• Written footprints: These can be online bread crumbs like old blog posts, LinkedIn updates, or video transcripts. For character, this might look more like dialogue snippets or descriptions.

• Behavioral indicators: Habits tell a story. Think app usage patterns, frequent pain points (or questions asked), as well as desired outcomes. These build a sense of priorities and pacing.

• Direct interactions or previous feedback: Sales calls, webinars, or support conversation transcripts are gold. They reveal tone under pressure and what someone values.

• Personality profiles: Do you have access to a Myers-Briggs, ESFJ or the like? Provide those to set the foundation for both communication strategy and how your twin sees the world.

• Demographic data: I’d argue these are less important than the above (and more susceptible to skewing), but things like age and gender can be helpful baselines to provide.

Aim for relevance over volume. Start with 5–10 strong data examples across different mediums and circumstances. The more diverse the input, the more balanced the twin.

Step two: Run the digital twin creation prompt (with your collected data)

Once you’ve assembled your data, it’s time to build the twin. You’ll do this by giving ChatGPT a one-time prompt that introduces the personality and hands over the materials.