You have the idea but can't execute. The instinct is to organize first: outline, structure, plan. By the time you've formatted the thought, you've lost the thing you were trying to capture.
This happens to writers, designers, strategists. You sit down to work and spend the first hour trying to arrange what you already know into something presentable. The raw version was richer. The structured version is thinner. And you can feel the difference, even if you can't name it.
Structured prompts compress what makes your thinking yours. When you organize your ideas into bullet points before giving them to AI, you strip out the texture, the hesitation, the unexpected connections. The AI then works with your outline instead of your thinking.
Input Inversion reverses the direction. You dump raw: talk, ramble, contradict yourself, think out loud. The AI finds the structure inside what you said. The ideas that come back are yours because the raw material was yours. The AI did organization, not ideation.
Experiential / contrastive: do it wrong first, on purpose
You pick a real project you're working on right now. Something with stakes. Then you do what the industry trained you to do: write a careful, structured prompt. Outline your goals. Specify the format. Give the model clear instructions. Capture the output.
This week exists because telling someone "structured prompts produce thinner output" doesn't change behavior. You have to feel it. The structured attempt is the control sample. You'll need it in Week 2 when you hold the two outputs side by side and the difference is immediately visible in your own material, on your own project.
The pedagogy note: we start here because in January 2023, this is exactly where I started. "Help me develop the rules." "Write out the ruleset." That instinct to organize for the model before you speak is so deep that skipping past it doesn't work. You have to do it, capture the result, and keep going.
Deliverable: A structured prompt and its output for one real project. Saved as your baseline.
Experiential / contrastive: dump raw, compare side by side
Same project. Different approach. The course provides an interview scaffold: a set of open questions designed to get you talking, not typing. You think out loud about the same project you prompted in Week 1. Messy. Unfiltered. False starts, contradictions, mid-sentence corrections. You talk through what you're building, what you're stuck on, what you actually care about. Voice notes, dictation, a conversation with the model where you're not performing, you're just thinking.
Then you feed that raw material to the AI and ask it to find the structure. Extract the key ideas. Identify the throughline. Organize what's there.
Then you put the two outputs next to each other: the Week 1 structured-prompt output and the Week 2 inverted-input output. The difference is not subtle. The structured version is clean and generic. The inverted version has your fingerprints on it because the raw material had your fingerprints on it. The AI did organization, not ideation.
The barrier here is psychological, not technical. The industry trained practitioners to perform for the model: clean prompts, specific instructions, defined formats. In November 2024, I wrote: "I don't really want this to be a session where you ask me questions. I really was just trying to have a brain dump." That sentence took two years to arrive at. Unlearning the performance is harder than learning the tools.
Deliverable: A raw brain dump (recorded or dictated) and its AI-extracted structure, held side by side against the Week 1 baseline.
Independent practice: build your own repeatable workflow
You have the evidence from Weeks 1 and 2. Now you build the practice you'll use after the course ends.
First, you learn to recognize the moment you start performing. There's a specific feeling when you shift from thinking to formatting. You catch yourself outlining instead of talking. You start editing before the thought is finished. That shift is the signal. When you feel it, you switch modes: stop typing, start talking.
Second, you build a corpus strategy. Voice notes on the commute. Dictated observations after a meeting. A conversation session where you think through a problem out loud instead of staring at a blank document. The course provides templates for each format, but the practice has to fit your actual life. A single dictated session where you speak freely about your work produces richer material than a carefully structured brief.
Third, you design your own input inversion pipeline: the specific sequence of dump, extract, structure that works for your projects and your tools. The pipeline is yours. The course gives you the components. You wire them together based on what you learned about your own process in Weeks 1 and 2.
Deliverable: A working input practice (corpus strategy, pipeline, and tooling) with before/after evidence from your own project.
Same person. Same project. One structured prompt. One brain dump.
January 2023: a designer asks AI to help build a tabletop game. He writes a clean, structured instruction. December 2024: the same designer, same project. He talks. Messy, unfiltered, mid-sentence corrections. Two inputs. Two outputs. The variable was not the model.
"Help me develop the rules and concept for this game."
"Sure, developing the rules and concept for your 'New City' tabletop RPG will require a significant amount of time and effort, but here are some steps you can take: 1. Define the game's goals and objectives. 2. Identify the game's core mechanics. 3. Establish the game's theme and setting..."
"I don't think there are any permanent structures in New City. maybe like decrepit buildings from whatever City used to be where the starport is, but I think mostly it's tents futuristic tents with cables and trusses... grungy and grimy."
"The Starport: the planet's only point of entry and exit, exclusively accessible to the wealthy. A sharp boundary separates the starport from the rest of the city. Core of New City: chaotic and alive, narrow winding paths crammed between shelters and makeshift stalls."
This course applies the research published in the Input Inversion whitepaper. Why unstructured human thinking produces better AI output than structured prompts.
This course is in development. $499 per course. Payment plans available on all courses. Foundations (Course 01) is the prerequisite. See the full curriculum.
$499. Requires Foundations. Payment plans available.