Billion-dollar organizations of one | Dabble.AI #7

The Dabble.AI Project #7 - Exploring autonomous business models

I haven't focused much time on the dabble.ai project since my last post on December 29, 2023. But, I have been thinking about it a lot.

I took this picture over the holiday this past year. It’s in Breckenridge Colorado.

My goal for this project is to figure out how to leverage AI to plan, build, launch, and grow a profitable company that has only one employee - me. Even though my company Dabble Lab has incredibly talented people that could help me out, I want to do everything myself. I'm doing this to hopefully prove it's possible.

I'm convinced that in the not so distant future, anyone will be able to easily use AI to plan, build, and launch companies that are also mostly run by AI. The human role will be primarily focused on vision, strategy, and oversight. In recent history, the norm has been to work with, or for someone else. That's changing. It's been changing for some time. But AI is going to impact that change in a big way. Like, I fully expect that we'll see billion dollar companies with only one employee in the next 5-10 years. Maybe sooner.

The general trajectory is that AI will continue to improve, enabling humans to do more and more, until almost everything can be done by a single person. I'm not saying that's going to happen overnight. Or that it will happen for every type of business. But it's going to happen. It's just a question of how long it takes.

Who will create the first billion-dollar companies with no employees?

I don't know. But my guess is that it won't be the type of person we might expect. The idea of using AI to create a billion-dollar startup seems like something that is going to take a lot of technical expertise. Maybe not. From my perspective, a deep understanding of technology isn't all that important. What's important is understanding the problem you're trying to solve. And then being creative enough to figure out how to use AI to solve it.

I'm a technical person. I've been writing code and building software for over 30 years and working with AI for over 10 years. But I'm realizing that my past experience isn't all that helpful for this project. In fact, at times, it's a hindrance. Because the ways I would normally approach a problem are not always the best approach when AI is involved. I catch myself making three left turns to go right. Said another way, I find myself over complicating things. I'm also finding that too often I underestimate what AI can do. Here's an example.

As I've shared in previous posts, I'm working through how I can use AI to write a book about Game Theory. It's not as simple as just saying to ChatGPT - "Write me a book about game theory". I've tried that. It doesn't work. But if I break the process down into smaller steps, it does work. For example, I can ask ChatGPT to write an outline for the book. Then an outline for each chapter with bullet points for topics that should be covered. Then I can have ChatGPT write content for each bullet point. 

This is a lot of manual prompting. But it works. And it works well. So, of course, as a software guy, my mind goes to writing code to automate the process of submitting all the prompts. But I'm trying to be mindful about my goals for this project - to have AI to do everything. So, I'm resisting the temptation to write code. I could just suck it up and submit all the prompts manually. But again, my goal is to have AI do as much as possible. Plus, that's way more time than I want to spend. So, I'm forcing myself to ask how I can get AI to do it - anytime I’m faced with something time consuming or that requires a lot of human expertise. 

Of course, my first thought was to just have AI write the code to automate the process of submitting all the prompts. That’s pretty easy to do. But then I started wondering if there was a way to get ChatGPT to prompt itself. I've dabbled with projects like AutoGPT and BabyAGI that are using this general approach. But those tools are code dependent and require a reasonable amount of technical skills to get them up and running. What I wanted was just a simple way to get ChatGPT to walk through a prompting process for generating a book. Without any code or external tooling.

I got it working. And it works surprisingly well. I'll share details in a future post. But the point is, I'm having to force myself to think differently. To almost unlearn years of training and conditioning. It's not been easy for me. But I'm sure it will feel natural for some. That's why I'm speculating that the first billion-dollar companies with no employees could likely be created by people that don't have a technical background. Maybe people without a lot of business experience either.

That’s it for now. Thanks for reading and stay tuned. The plan for my next post is to outline the book generation process.