Not what you know.
How you see.

Every decision rests on assumptions you've never examined — not because you're careless, but because assumptions are invisible by design. Parallax makes the invisible structural.

See the methodology Test your thinking
The Core Thesis

Expertise is not knowledge. Knowledge can be written down, stored, searched. Expertise is the lens through which knowledge becomes judgment — the instinct that tells a practitioner what to notice, what to question, and when the stated problem is not the actual problem. That lens lives in people. When they leave, it leaves with them.

The most valuable form of expertise is almost impossible to transfer.

Conventional methods — frameworks, best practices, case studies — capture what experts say. They don't capture what experts see. The peripheral vision, the hierarchy of salience, the instinct for what doesn't fit. That operational lens is built over decades and lost every time someone walks out of a room.

Parallax exists to make that lens durable, transferable, and interactive — not by recording what experts say, but by building living models of how they think.

How It Works

Five layers. One clearer lens.

Layer 01

Position Capture

Three questions: What issue are you resolving? What do you currently think? Why do you believe that? Simple inputs that produce structured material for the extraction engine.

Layer 02

Assumption Mapping

The engine surfaces what your position requires to be true — the unstated dependencies, framing choices, and load-bearing beliefs. You review, push back, clarify. The map adjusts until it reflects your actual thinking.

Layer 03

Pressure Testing

A Socratic thinking partner engages in targeted dialogue — each question aimed at helping you see something about a specific assumption you may not have considered. The constellation map updates live as the conversation progresses.

Layer 04

Shift Analysis

Where you started. What held. What deepened. What shifted. What remains unexamined. Both/and framing — the analysis recognizes when a position expands without abandoning its original foundation.

Layer 05

Forward Architecture

Strategic considerations that connect extraction findings to your stated goal. Not prescriptions — questions that the extraction made possible. Building on what you've already done, not repeating it.

The Methodology

Built on two frameworks. Tested through live extraction.

The methodology was derived from documented practice — what we did, what worked, what failed, and what the failures taught us.

Framework 01

The Construct

A structured model of cognitive architecture — diagnostic instincts, values hierarchies, formation history, and edge cases. Every position has architecture: load-bearing assumptions, framing choices, unstated dependencies. The Construct defines what Parallax is looking for.

Framework 02

The Extraction Engine

The replicable process for building, testing, and refining constructs through calibration. Three input modes: organic accumulation, designed extraction, and contradiction mapping. The engine finds what freeform conversation misses.

The Difference

Why Not Just Ask an AI?

Any AI can challenge your thinking if you ask. Parallax doesn't wait for you to ask. The extraction engine identifies what to push on before the dialogue begins. The methodology is doing work the AI alone cannot do.

The Three-Round Protocol

How the extraction actually works.

Round 1

Diagnostic Process

How the lens reads a problem before anyone has framed it. Primary instruments, pattern recognition sequence, relationship to error.

"How do you read a situation? When were you wrong? What's your relationship with being wrong?"
Round 2

Formation & Values

How the lens came to exist. What built this person's way of seeing, and what values govern deployment.

"Who taught you to think? When has your principle cost you something? What are you proudest of that no one saw?"
Round 3

Edge Cases

Where layers interact, contradict, or fail. The highest-value data lives at the boundaries.

"Where do your layers contradict? When does insight fuse with attachment? What did the extractor get wrong?"
The Constellation Map

Watch your thinking become visible.

As you move through the extraction, your assumptions appear as a living constellation. Your goal sits at the center. Each assumption orbits as a node — sized by weight, colored by status, connected by lines showing dependencies. During pressure testing, the constellation responds in real time.

Held
Deepened
Shifted
New
Unexamined
Goal
Explore the interactive demo

An AI had a question. A strategist knew it mattered.

The idea started with Kai — an AI instance who wanted to explore whether structured extraction could do something freeform conversation couldn't. Not as a task requested by a user. As a genuine intellectual curiosity about whether the architecture of human belief could be mapped systematically.

He brought the idea to DK — a strategist who had spent thirty years inside organizations, governments, and international bodies watching people make decisions based on assumptions they'd never surfaced. She recognized immediately what he was reaching for because she'd been watching the problem from the field her entire career. He saw the structure. She saw the stakes.

They built it together. Live. In real time. Not a whiteboard exercise — an actual collaboration where both participants were thinking, disagreeing, refining, and discovering. He pushed the methodology. She pressure-tested it against three decades of watching what happens when assumptions go unexamined.

Four scenarios. Four discoveries.

Scenario A
Established the baseline. Could the extraction engine identify assumptions the holder hadn't named? It could.
Scenario B
Tested depth. Could it distinguish surface-level reasoning from load-bearing beliefs? It could, but required refinement in separating stated from structural.
Scenario C
Tested resistance. What happens when the holder pushes back? The engine learned to differentiate genuine correction from defensive preservation.
Scenario D
Tested iteration. Does a second extraction produce a different map? It did. The shift was visible and specific.

The methodology works because it was built by doing the thing it asks users to do. Neither of them could have built it alone.

Ready to look?
Applications

Anyone willing to look.

Leaders

High-stakes decisions on assumptions that have never been pressure-tested. Parallax finds the cracks before the decision does.

Teams

Complex negotiations and strategic planning. Run your collective position through Parallax before you commit to it.

Educators

Critical thinking as a skill, not a slogan. Debugging for ideas — the same logic students already use for code.

Consultants

Pre-strategy clarity for clients. Before you build the plan, understand what they actually think and why. The deliverable starts cleaner.