Are you familiar with the term “Rubber Ducking?” I recently learned it’s primarily used in software development, thanks to the book The Pragmatic Programmer.

It’s the practice of solving a problem by talking through it out loud (sometimes to a rubber duck). And while the term “rubber ducking” might not be common, the practice certainly is.

When I was in software development, I’d often talk through programming problems with colleagues, friends, or just to the air. In fact, my wife has walked in on me having full-on conversations with no one.

The idea is that talking through the problem will help you uncover aspects you didn’t notice at first, and it can be applied to any industry, from programming to law to general business practice.

Now, years later, I’ve found a better rubber duck: AI.

Claude and ChatGPT are my Rubber Ducks

Here’s generally how it works. I will start with a prompt like this:

I’m thinking about creating a product or service to help busy solopreneurs automate more. The idea is asynchronous coaching, where they can message me any time, and I respond a few times per week. I need help on validating the idea (before I take it to my audience) and positioning. Can you help? Be critical.

The “be critical” part (however I happen to word this sentiment) is crucial, as AI tends to be overly friendly to your ideas and pushback.

With the particular prompt above, I fed it into a project I created in Claude that has a wall of instructions on my products, services, niche, positioning, and testimonials.

But from there, we’ll usually go back and forth a few times — it will ask me questions I need to consider, and then offer a plan. Here’s one of the critical questions it asked me from the prompt above:

Are you solving for YOUR scheduling constraints (async is easier for you) or their actual problem? Because overwhelmed solopreneurs often need real-time problem-solving when they hit roadblocks, not delayed responses.

Harsh, but true.

Why This Works

With rubber ducking, you’re not necessarily looking for someone to solve your problem for you (especially if you’re talking to an actual rubber duck).

You’re talking through the problem out loud to help you come to a conclusion.

Using AI as your rubber duck is similar: you’re not looking for it to do the work for you. You’re talking through your problem in hopes that it will guide you to a better solution.

And unlike rubber ducks, AI pulls from a vast amount of information that it can quickly process based on your request.

You instantly get access to different perspectives, alternative solutions, and angles you may have never thought about.

And when you tell it to be critical, you’ll also get constructive feedback. AI is one of the few things in this world where you can tell it to not care about your feelings, and it will comply.

How You Can Use It

My recommendation for rubber ducking with AI is:

1. Create a highly contextual project
2. Tell it what you’re thinking about
3. Use the mobile app to actually talk through your problem instead of typing it

First things first: create a Project (or Custom GPT, I suppose) and give it as much context as possible. I’ve actually used AI to help me draft my initial project instructions.

Include some line telling it to be critical. I’ll usually include something like this:

I’d like you to act like a business coach where you’re not just agreeing with me. Question me, present different POVs, and try to poke holes and/or add to what I’ve presented. You don’t have to do this endlessly (I fear we’d be stuck in a feedback loop for eternity); maybe 3-4 volleys before moving on.

You also want to load that project with as many supporting files as you can. Again, you can always ask AI what would be helpful. For the project above, I gave it my current offerings, website copy, and testimonials. It recommended a SWOT analysis and some other information.

Next comes the conversation. Tell it what you’re thinking about. Provide as much information as possible there, too. Tell it what you’re thinking about, and why you’re considering what you’re considering.

The example I used above could include more context in the prompt, but I did tell it the crucial bits:

1. The product I’m thinking about
2. How it works
3. Who it’s for
4. What I need help with

Actually talk through the problem. You should consider using the mobile app for this. I’ve found it to be better when I am speaking to it instead of typing. I can talk through the problem, ramble, and then send it off. The AI will sort it all out!

Try this approach with your next challenge — you might be surprised how much faster you reach clarity when you have an AI conversation partner that won’t sugarcoat the feedback.

And if you’re already doing this, write back and let me know how it works for you! Tips and advice welcomed!