Stage Overview

The core challenge: Knowing what you want before you ask.

The hidden difficulty: Most people at Stage 1 don't know they're at Stage 1. The problem is invisible until AI surfaces it — and then it gets misattributed.

The CDPID reframe: You don't have to have intent fully formed before you start. Intent can emerge through the collaboration. Stage 1 development means learning both how to clarify intent before sessions and how to develop it through them.

What This Stage Actually Is

Here's the experience at Stage 1 that almost everyone recognizes once they hear it described:

You think you know what you want. You ask. You get output. The output is... fine. But not quite right. You ask again with slight adjustments. Different output. Still off. You try a third variation. Closer, but still not what you actually wanted.

The natural interpretation: AI is limited. It doesn't understand nuance. It's not smart enough for this particular use case.

That interpretation is almost always wrong.

AI is doing exactly what you told it. The problem isn't AI's capability — it's that the representation of your intent that you gave AI wasn't precise enough to be translated into what you actually wanted. The gap isn't AI's limitation. It's an intent-clarity gap.

Why this is invisible: Your internal experience of your intent feels complete. You know what you want — or you believe you do. The discovery that you don't know as precisely as you thought only happens when you try to communicate it to something that has to work with the representation you provide.

AI is, among other things, a precision mirror for the clarity of your thinking. Sessions that produce poor output are often diagnostic — they're showing you something about the clarity of the intent you brought, not the capability of the tool.

The CDPID Insight at Stage 1

CDPID — Collaborative Driven Progressive Intent Discovery (DOI: 10.2139/ssrn.6278438) — provides a framework that reframes Stage 1 entirely.

The central finding: intent isn't something you bring to an AI session fully formed. Intent is something that emerges through the collaboration. Neither the human nor AI has complete understanding at the session's start. Both contribute to developing it through the process.

This means Stage 1 has two distinct development tracks:

Track 1 — Pre-session intent clarification: Developing the ability to know what you mean precisely enough before you open AI to begin working. Getting sharper at articulating requests, at knowing what you want to accomplish and why, at identifying the real question underneath the surface question.

Track 2 — Collaborative intent development: Learning to use the session itself to develop intent. Rather than trying to have everything figured out before you start, you enter with what you have and use the conversation to develop clarity progressively. This is CDPID in direct practice — intent developing through collaboration rather than arriving pre-formed.

Both tracks are valid. The most capable Stage 1 practitioners use both: they bring more clarity to each session than they did in the last one, and they know how to use the session itself to develop clarity when they're starting without it.

What You Can Do at This Stage

When Stage 1 is genuinely developed, you can:

• Articulate what you want from a session before you begin it, specifically enough that the output is recognizable when you get it
• Identify the difference between "I know what I want but can't express it" and "I'm not sure what I want yet" — and respond differently to each
• Use pre-session intent writing to sharpen vague requests into precise ones before asking
• Recognize when output misses your intent and begin diagnosing where the gap came from
• Use the session collaboratively to develop intent when you're starting with something unclear
• Catch the "fine but not right" response and use it as a diagnostic rather than a frustration

These don't feel like big capabilities until you've worked with people who don't have them. The difference in session quality — and in the time spent on iteration cycles — is significant.

The Ceiling at This Stage

The Stage 1 ceiling: every session starts from roughly the same place.

Even when your intent is clear and well-articulated, there's a baseline level of context-rebuilding that happens every session. You re-explain your situation, your preferences, your relevant background. AI works with what you give it in that session, and produces output calibrated to that session's context.

That ceiling isn't about intent clarity — it's about context architecture. Building the infrastructure for sessions to start from an informed baseline rather than from scratch is Stage 3 work.

But there's also a subtler Stage 1 ceiling that shows up before Stage 3: even with perfectly clear intent, your ability to communicate that intent to AI with precision is still developing. That's Stage 2 territory — the cognitive translation capability that closes the gap between what you mean and what AI can work with.

How to Develop Through This Stage

Pre-session intent writing. Before you open AI, write out in plain language: what do you want? What will you do with the output? What makes this output good vs. not good? Read it back critically. Is this specific enough for someone else to work from? That "someone else" test is useful — if you couldn't hand this brief to a smart colleague and get back what you want, it's not specific enough for AI either.

The "why" test. For any request, be able to articulate: why do I want this? What's the real need underneath the surface request? This often reveals that the request you were about to make is a solution to a problem, and the actual problem you want solved is something slightly different.

The "real question" excavation. Practice finding the question underneath the question. What you type first is often not the actual question — it's the first approximation. What are you actually trying to understand, decide, or produce?

CDPID practice. Enter sessions with what you have, not with what you wish you had. Let intent develop through the collaboration. Notice when the session produces something that clarifies what you actually wanted (more useful than the output itself) — write that down before the session ends.

Intent-output gap analysis. When output misses, don't just retry. Diagnose. Was the gap because: (a) you didn't know what you wanted clearly enough (Stage 1 gap), (b) you knew but couldn't represent it (Stage 2 gap), or (c) AI didn't have the context to work from (Stage 3 gap)? Getting precise about which gap you're in tells you what to develop.

Signals You're Ready for Stage 2

You're developing past Stage 1 when:

• Your sessions consistently produce output that is recognizably what you wanted (even if imprecise in other ways)
• You can reliably reproduce good session outcomes — you understand what produced them
• The gap you experience in sessions is less "I didn't know what I wanted" and more "I know what I want but AI produces something adjacent to it"
• Pre-session intent writing feels natural and speeds up your sessions rather than slowing them down
• You've started to sense that the problem in missed outputs is about communication precision, not intent clarity

That shift — from intent clarity gaps to communication precision gaps — is Stage 2 territory.

Community Resources for Stage 1

Stage 1 Group — Intent Clarification Share your intent-clarity challenges, bring real session diagnostics, and learn from others at the same stage. The most valuable discussion here: bringing a session that didn't produce what you wanted and working through where the intent gap was.

Resources in this stage:

• Intent clarification exercises and diagnostic frameworks
• Pre-session intent writing templates
• CDPID introduction (what collaborative intent development looks like in practice)
• The AI Mastery Toolkit — orient to it now to see where Stage 3 leads

The Healthy Dimension

Healthy Stage 1 development: Using AI as a mirror that surfaces unclear thinking, not a tool that compensates for it. Treating missed outputs as diagnostic information about your own intent clarity. Getting progressively better at knowing what you mean before you ask, and at developing clarity through the collaboration when you don't.

What to watch for: The Stage 1 failure mode isn't dramatic — it's subtle. It's using more prompts, more retries, more tool-switching rather than developing the underlying clarity. It's interpreting inconsistent results as AI limitations and collecting workarounds rather than building capability. It's staying in a transaction relationship with AI (ask → get output → repeat) when the developmental move is toward something generative.

The measure of Stage 1 development isn't output quality — it's your relationship to unclear intent. Are you getting better at knowing what you want? At using collaboration to discover what you want? At catching the moment when you don't know and using that as a signal rather than a frustration?

That development is what carries forward into every subsequent stage.