v0.1 · pressure-test  ·  not public, not indexed  ·  an active hypothesis built to be challenged. Tell me where it's wrong.
Active hypothesis · v0.1 · the overhead view

When AI collapses the cost of building, coherence becomes the scarce resource.

Parallax is a viewpoint name for the overhead view of a coherence stack: a spine of composable products and frameworks, organized on two structural axes and one production chain, composing along one loop (Notice, Spec, Execute, Observe), governed by the Intent operating model. It is not a product, not a container, and not a methodology book. It is the surface from which the relationships between the parts become legible — and challengeable.

the coherent system viewpoint A viewpoint B altitude separation

Two vantage points, one distant object: depth appears when separated viewpoints are held together. That is what this system does to its own work.

Why it exists

AI changed the cost curve, not the destination.

Everything downstream of the model is now cheap — the implementation, the typing, the boilerplate. Everything upstream of the model — the noticing, the framing, the deciding — is where the leverage moved. Teams that keep optimizing for the old constraint (coordinating human labor through tickets and sprints) build overhead around the wrong bottleneck.

The founding reframe is one move: from a coordination tax to a coordination investment. Agile ceremony evaporates after each sprint — it is paid down, repeatedly. Coherence artifacts compound — they are paid forward, once. That single shift is the argument the discipline rests on.

Coordination tax
Standups, story points, sprint rituals. Re-paid every cycle. Evaporates on completion. Scales with headcount.
Coordination investment
Compiled knowledge, specs-as-contracts, preserved dissent, traceable decisions. Paid forward once. Compounds.
What Parallax is

A viewpoint, not a container.

Parallax sits above the stack as the shareable narrative surface. It composes the overhead view from the products' state — it does not own or contain them (sibling-over-parent-child is the architectural law here).

It is

  • The overhead view — altitude model + the relationships across the stack
  • A surface that publishes from the products below, never wraps them
  • Brand + narrative architecture — a way to read the system, not run it
  • An honest hypothesis under pressure test

It is not

  • The discipline — that is Coherence Engineering, held internal until the gates open
  • The execution framework — that is Intent, at a different pipeline position
  • A container or parent of the other products
  • A category — not yet. It earns the name by what it produces, not what it claims
The stack as proof

It built itself. That proves it runs.

This is not a theory deck. It is a working system of composable products along one loop, and the loop ran on its own construction. Honest limit, adjudicated on the record: self-construction evidences that the system runs; it does not evidence that it generalizes. Agreement without displacement is cache. Generalization is exactly what the gates below exist to earn.

Notice Spec Execute Observe double-loop: Observe updates the domain model, not just the execution
Notice
Intent signals · Witness events · Pulse watch-list · Cortège fetch fabric
Spec
Voices dissent panels · Cast identity · Crucible synthesis · Topography plan-state · Warp altitude tensioning
Execute
Forge composition · Fieldbook · studio-control · org-design-tooling
Observe
library-index · Loom session-state · Conduit cross-surface · Throughline trace · Gauntlet disconfirmation

Intent is the reference implementation — the execution loop, proven first, with its own published site. The rest of the spine is the same discipline applied at other positions. That is the doorway: you arrive from Intent, and this is the system Intent is one instance of.

Lineage — nothing here is invented

The credit travels with the ancestors.

This is a synthesis, not a novel method. Each load-bearing idea is borrowed on purpose, and named — because the credibility comes with the name.

Marty CaganEmpowered teams — give the problem, not the solution. AI widens the gap.
Jeff PattonShared understanding is the artifact; tickets are residue.
Teresa TorresContinuous discovery — keep the opportunity tree alive, don't freeze it.
April DunfordPositioning — what category, against what alternative, and why best.
John BoydYou can't get coherent perception from inside — inject external reference points (→ Voices).
Chris ArgyrisDouble-loop — question the assumptions, not just optimize execution.
Andrej KarpathyThree layers — raw sources / compiled knowledge / running code. Compile, don't retrieve.
Nate B JonesThe three-engineering progression — prompt → context → intent engineering.
Where it's unproven

This is a hypothesis. Here is where to swing.

The honest standing of the claim

This is not yet a category; it is a hypothesis under pressure test. The discipline name — Coherence Engineering — is held internal on purpose, and only earns public use once three external-articulation gates clear. Until then the work speaks in its sibling vocabulary: prompt, context, and harness engineering.

And the honest market: the serviceable addressable audience today is roughly one — me. That is exactly why this surface exists. A worldview cannot advance past theory, or past an audience of its author, unless its view is expressable, navigable, and challengeable. So: navigate it, and challenge it.

Gate 1 · open The Voices-judged coherence-debt register resolves to zero.
Gate 2 · open The CON-COH machine-assertable contracts pass against the corpus.
Gate 3 · in test "Parallax" reads as depth-through-viewpoints to an external reader without explanation. (This page is the instrument; the first peer read is the test.)

Latest evidence the challenge is real: the altitude gate's first recorded runs (2026-07-19) failed the portfolio's own artifacts: the strategic root thread at 0.40 tension, the positioning doc at 0.00. The instrument refuses what it cannot verify, and the findings are open work, on the record.

Shipped: Intent loop, Voices, Cast, library-index, Warp altitude gates (live verdicts), Gauntlet disconfirmation Hypothesis: coherence-as-discipline generalizing beyond its author Open: is this a category, or just my operating system?

The sharpest open question, stated plainly for the reader most likely to ask it: is the surface area too large? A product spine this wide is either coherence operationalized, or sprawl with a viewpoint name on top. If you'd implode part of it, that's the conversation this page is for.

Who it serves

Built for one, opened outward deliberately.

Now
The author
A solo practitioner at the scale of a small consultancy. The system is the second practitioner he doesn't hire.
Inside the tent
Registered peers
Architecture/AI and product-academy peers, in the Cast registry — patterns and challenge flow both ways.
Downstream
Engagement teams
Receive the methodology's rendered outputs — a polished artifact, not an unfamiliar process to adopt.
Gated
The published audience
The "coherence engineering as a category" reader — behind the three articulation gates above.
Where Parallax stands — and how to go deeper

Against what alternative.

Dunford's question: what are the real alternatives a collaborator has today, and why does none of them hold this position?

Fragmented consulting frameworks — disconnected playbooks for discovery, delivery, strategy. No altitude model, no throughline between them.
Agile transformation services — execution-posture work (how we work) that rarely touches the altitude problem (are we even at the right altitude?).
Narrative strategy firms — they work on how you describe the work, not on the coherence of the work itself.

None holds altitude + coherence + narrative simultaneously. That intersection is the position — claimed as a hypothesis, not yet as a category.

The architecture
Every product, framework, and seam, with paths and contract authorities. Core/ECOSYSTEM-ARCHITECTURE-2026-05-20.md
The why, in full
The unifying narrative for the stack. Core/ECOSYSTEM-MARKETING-NARRATIVE-2026-05-20.md
The discipline
Coherence Engineering, hypothesis-in-refinement. Core/frameworks/coherence-engineering/DEFINITION.md
The reference implementation
Intent, the execution loop, proven first, with its own site. theparlor.github.io/intent-site