Kieran Klaassen

I'm walking down the street when an idea hits me. It feels brilliant, complex, full of potential. In the past, I might have scrambled for a notebook, fumbled with my phone's notes app, or, worse, trusted my memory to hold onto it. By the time I could capture the thought properly, it had usually faded into a fuzzy outline of something that once seemed profound.

Lately, I've been using a simple method to capture and develop those fleeting ideas. Voice recording has been around for ages. Pair it with AI, though, and it becomes far more useful for anyone who has watched a good idea slip away.

It starts with a voice memo.

The Elusive Nature of Flow

Psychologists call that state of deep focus "flow": ideas come freely, and time seems to disappear. Anyone who makes things knows how unpredictable it can be. Flow keeps its own schedule and often shows up at the least convenient time.

Psychologist Mihaly Csikszentmihalyi, who pioneered the scientific study of flow, described it as "a state in which people are so involved in an activity that nothing else seems to matter." His research showed that people are happiest and most creative when they're in this state. He also noted its elusive nature: flow can be cultivated, but not forced.

The harder part is keeping what comes out of flow. How many times have you had an idea in the shower, while driving, or just before falling asleep? By the time you're able to act, the thought may still be there, but its original clarity and energy are gone.

Voice Memos: A Simple Yet Powerful Tool

Voice memos are hardly new; portable recorders have existed for decades. Smartphones have simply put one within reach at almost any moment. You don't need to type or find a flat surface. You hit record and speak.

But here's the rub: voice memos were often where good ideas went to die. We'd record our thoughts in a burst of inspiration, then never listen to them again. Transcribing was a chore, and even after doing it, we'd be left with raw, unstructured text. Turning the recording into something useful took too much effort.

Sidenote: There's an interesting philosophical angle here. Philosophers Andy Clark and David Chalmers proposed the "extended mind," arguing that external tools can function as part of our cognitive processes. Seen this way, voice memos do more than capture thoughts. They extend how we think and create beyond the confines of our biological brains. The idea becomes even more provocative once AI enters the process.

AI changes what happens after you press stop.

The AI-Powered Transformation

Those neglected voice memos can now become workable material. Here's how I use them:

Capture: Record your thoughts with your phone's voice memo app. Don't worry about structure or polish. Let the ideas come.
Transcribe: Use a transcription service like Whisper to convert the audio into text. What used to be the bottleneck can now happen in minutes with remarkable accuracy.
Expand and refine: Feed the transcript into a language model like ChatGPT. Ask it to structure the ideas, expand key points, or critique the argument from another perspective.

The result is a fleeting idea captured and developed into a usable draft with less effort.

More important, the process lets you lean into the moment. While you're in flow, you can keep generating ideas and leave the structure for later.

This is a form of cognitive offloading: an external tool handles part of the mental work. AI can take a rough sketch and help turn it into a draft. That leaves us more attention for generating ideas while the machine helps with the structure and refinement that can bog us down.

The approach respects how fragile flow can be. We can stay with the thought instead of interrupting it to worry about structure, knowing we have a way to shape the raw material later.

Crafting Effective Voice Memos

AI can help with the later stages, but the output still depends on what you record. A few habits make those recordings easier to work with:

Embrace stream of consciousness: Don't self-edit as you speak. Let the thoughts come; AI can help organize them later.
Provide context: Briefly explain the background or motivation behind the idea. The model can't infer what you never said.
Ask questions: Say where you're uncertain or what you want to explore further. You can ask the AI to expand on those points later, then verify what it gives you.
Suggest structure: If you have a rough idea of how to organize the material, say so. For example, "I think this could be broken down into three main points..."
Include prompts for the AI: As you record, say how you might want the model to work with the material. You could add, "It would be interesting to look at this from both a psychological and nutritional perspective," or "I'd like to explore some counterarguments to this idea."

The goal is to keep enough of the original thought that you can return to it later, so forget perfection and get the core ideas down while they're fresh.

This approach also uses metacognition—thinking about thinking. Recording with later processing in mind means we're already considering how an idea could be developed, even before we sit down to write.

It encourages us to consider different angles and possible expansions while the idea is still forming. Because the execution can wait, we can stay in a generative mindset. The recording holds both the in-the-moment thought and a little guidance for the person—or model—who will shape it later.

Philosophical Perspectives on Creativity and Technology

The method raises a larger question: does AI expand our creative abilities, or make us too reliant on an artificial aid?

In 1967, John Culkin, a friend and colleague of Marshall McLuhan, wrote, "We shape our tools and thereafter they shape us." The line feels especially relevant to AI-assisted creativity. As these tools become part of the process, they may reshape how we think and create.

The optimistic view, represented by Clark and Chalmers, treats this as an extension of human cognition. Their "extended mind" thesis suggests that an external tool can become so integrated with our cognitive processes that it functions as part of the mind. From that perspective, AI tools can expand what our minds do, much as writing expanded our ability to store and transmit knowledge.

AI then becomes a collaborator in the creative process, helping us explore connections that might otherwise be difficult to see. Human creativity still directs the work, but it has another instrument at hand.

The counterargument matters, too. If we always hand structure and development to AI, do those abilities begin to atrophy? And if many creators use similar models in similar ways, does the work start to sound the same?

There are no easy answers. The sensible approach may be to use AI while continuing to practice the skills we're tempted to hand over. The initial spark and the connection between disparate ideas still come from the person using the tool. AI can help develop them, but we decide where it points.

Every time we fold AI into a creative habit, we also make a small choice about what human-AI collaboration will look like.

Conclusion

Voice memos and AI give us a practical path from fleeting inspiration to structured material. The voice memo preserves the thought; transcription makes it workable; and the language model can help shape it.

The method also asks us to pay attention to what we delegate, especially the difference between capturing an idea and letting a model determine what it means.

Used deliberately, these tools can make the gap between inspiration and finished work smaller without taking authorship out of our hands.

The next time an idea arrives while you're walking, grab your phone and hit record. Keep the strange detail, the half-formed connection, and the energy in your voice. Those are the parts worth carrying back to the desk.