Typing is starting to feel obsolete for capturing complex ideas. We often think of prompt engineering as a precise, text-based discipline requiring perfect syntax, but that might be slowing us down. I recently stumbled upon a fascinating post from a forward-thinking AI professional who is completely rethinking how we interact with Large Language Models. This innovator describes a method of “prompt engineering via voice notes” that transforms the chaotic noise of our minds into structured gold. Instead of sitting at a keyboard, the creator walks around recording long, messy audio files and lets AI do the heavy lifting.
⚙️ The Mechanism: From Audio to Architecture
The core concept here is untethering the brain from the fingers. The author explains that the voice is the most efficient vessel for raw thought because typing inevitably acts as a filter. When we sit to type, we self-edit; we worry about grammar, structure, and flow before the idea is fully formed, which often kills the creative spark. By recording “long rambles,” the original poster captures the nuance, the excitement, and the “noise” that usually gets lost in translation.
They use tools like Gemini to transcribe the audio instantly, but the process doesn’t stop at simple transcription. The expert feeds this raw, unstructured transcript into the LLM with instructions to parse the stream of consciousness. The model acts as an intelligent structural engineer, taking the disorganized audio and converting it into high-quality prompts, blog posts, or solutions. It bridges the gap between brain speed and typing speed, effectively using the LLM to organize the creator’s thoughts better than they could in real-time.
📌 Maximizing Information Fidelity
One of the most compelling points the expert makes is about the richness of the input. When you type a prompt, you tend to be concise, often stripping away the context that the AI actually needs to understand your intent. The LinkedIn user emphasizes that voice notes capture thoughts in their “original form,” preserving the emotional weight and subtle connections between ideas.
Think about how often you have a brilliant, complex idea while driving or walking, only to lose half of it by the time you open a notes app. By rambling into a recording, the creator ensures that every tangent and qualification is preserved. The AI can then look at the totality of that information to discern the core objective. This results in a higher fidelity output because the model has more data points to work with, allowing it to generate solutions that are surprisingly aligned with the user’s deeper goals.
💡 The Art of the “Messy” Input
We usually teach people to be precise with AI, but this industry pro argues for the opposite approach during the ideation phase. The post highlights that raw audio isn’t the final product—it is merely the raw material. By feeding a “messy” stream of thought into the system, the author allows the AI to perform the heavy cognitive load of structuring.
This is a massive productivity unlock. Instead of spending twenty minutes crafting the perfect paragraph, the creator spends two minutes talking through the problem out loud. The LLM cleans up the “ums,” “ahs,” and irrelevant sidebars, extracting the diamonds from the rough. This shifts the human burden from “organization” to pure “creation.” It turns the user’s rambling into a superpower, generating training topics and posts instantly without the friction of a blinking cursor.
🚀 Mobility as a Creative Catalyst
There is a distinct advantage to untethering yourself from a desk which goes beyond just saving time. The LinkedIn user jokes about looking like a “crazy man in black” wandering around Singapore talking to a phone, but there is genius in this mobility. We know that a change of scenery often triggers different neural pathways than staring at a static screen.
By combining physical movement with voice-based prompting, the creator is likely accessing a different level of creativity known as the “shower thought” effect. It turns the entire world into a workspace. You aren’t limited to where your laptop is; you can engineer complex prompts while waiting for coffee or walking through a park. This method allows the environment to influence the thinking process, leading to more dynamic and unexpected ideas than you would get from a sedentary office session.
Nuances and Potential Hurdles
While this sounds incredible, there is a learning curve to effective “rambling.” You have to get comfortable hearing your own voice and trusting the AI to sift through the nonsense without micromanaging it. Privacy is another major factor to consider; you probably shouldn’t be dictating sensitive proprietary data while walking down a crowded street. Additionally, the quality of the output depends heavily on the model’s context window and reasoning capabilities. If the rambling is too disjointed, even a powerful model might struggle to find the signal amidst the noise. It requires a bit of practice to learn how to verbalize your thoughts in a way that provides enough distinct hooks for the AI to grab onto.
This approach from 2025 is definitely a habit worth adopting for the future! I highly recommend checking out the full post to see exactly how the expert implements this workflow.