Coached Thinking

The COACH Framework

A five-practice scaffold that ensures AI augments student judgment rather than replacing it — with a structured artifact you can assess.

Why judgment formation requires a scaffold

When students use AI freely, the model's output tends to crowd out their own reasoning. Work looks finished, but the cognitive work that matters — noticing what's uncertain, weighing competing interpretations, owning a conclusion — often doesn't happen. Faculty lose the window into student thinking that assessment depends on.

The COACH framework changes the relationship between student and AI. Instead of asking the AI for answers, students engage it as a structured challenger: one that demands they articulate, examine, and defend their own reasoning before accepting or rejecting any input.

Each letter corresponds to a practice. Together they create a conversation arc that produces both a better final decision and a transparent record of how the student got there.

Cultivate Initial Judgment

Happens before any AI interaction

Before students see the AI's perspective on anything, they read the case and commit to an informed position. They describe what they believe the core problem is, what matters most, what assumptions they're carrying, and what they don't yet know.

This initial impression is saved and becomes the baseline for the rest of the session. The student can't revise it later — and both student and faculty can see how it compares to their final recommendation. That comparison is part of what you assess.

Why it matters: Without a committed starting point, students drift toward whatever the AI produces. The impression step gives them a stake in their own thinking before any external input enters.

Orient Attention

First coaching conversation

In the first AI conversation phase, the system prompt guides the AI to redirect student focus toward the dimensions of the problem they've underweighted. If a student's impression is heavy on financial considerations, the AI surfaces the people or strategic dimensions they've glossed over — without simply providing new conclusions.

The AI asks questions more than it answers them. The goal is reorientation, not correction. Students are pushed to notice what they weren't attending to.

Why it matters: Real judgment requires looking at a problem from multiple angles. This phase makes selective attention visible and creates an opportunity to broaden it before the student commits further.

Appraise Reasoning Quality

Second coaching conversation

In the second phase, the AI turns attention to the quality of the student's reasoning itself. It asks them to identify the assumptions behind their position, the evidence they're relying on, and the inferential leaps they've made — and to evaluate whether those hold up.

Students aren't just asked "is this right?" — they're asked to articulate why they believe what they believe, and then to examine that explanation critically.

Why it matters: Students who can produce a recommendation often can't explain the reasoning behind it. This phase surfaces the gap between conclusion and justification, which is where the actual learning tends to happen.

Challenge Before Choosing

Third coaching conversation

Before the student finalizes their recommendation, the AI steelmans the strongest competing position. It articulates the best case for the alternative and asks the student to engage with it seriously — not to dismiss it, but to explain why their position still holds (or to update if it doesn't).

Students are being asked to do something genuinely difficult: take the opposing view seriously enough to be uncertain about their own.

Why it matters: Premature closure is one of the most common failures in judgment. Confronting a genuinely good counterargument — rather than a strawman — makes the decision more robust and the reasoning more honest.

Hold to Account

Final accountability step

After submitting their final recommendation, students answer a set of accountability questions: What did they accept from the AI's input, and why? What did they reject or revise? What does their recommendation depend on? What's the strongest case against it? What remains uncertain?

These responses — alongside the full conversation transcript and the original impression — form the assessable artifact the faculty receives.

Why it matters: Accountability is what separates judgment from opinion. Requiring students to explain what they took and what they left creates ownership. It also gives you the evidence you need to assess whether genuine reasoning took place — not just whether the conclusion looks reasonable.

At a glance

Practice	When it happens	What the student does	What the AI does
C — Cultivate	Before AI interaction	Records initial judgment, key issues, assumptions, unknowns	Not yet involved
O — Orient	Phase 1	Engages questions about underweighted problem dimensions	Redirects attention without providing answers
A — Appraise	Phase 2	Examines assumptions, evidence, and inferential leaps	Probes reasoning quality; asks for justification
C — Challenge	Phase 3	Engages the strongest counterargument to their position	Steelmans the best alternative; asks for a response
H — Hold	After submission	Accounts for what was kept, changed, and why	Accountability questions are pre-configured per assignment

Bring the COACH framework to your course

Free to use. Set up in minutes. No LMS integration required.

Create a Free Account