Guided Intelligence
§5 · Intent Modeling

GI's single unified input format.

GI defines the internal mechanics of software creation, not how an organization decides what to build. The Intent is the boundary between upstream strategy and the GI pipeline.

§5.1 · What is an Intent?

A structured statement of the why and what.

An Intent expresses what the system should achieve without dictating how to do it. It isn't a user story, doesn't decompose into tasks, and doesn't predefine architecture.

The Planner defines the how.

An Intent includes
  • 01
    Desired outcome
    Business, user, or technical.
  • 02
    Constraints
    Time, risk tolerance, rollout strategy, UX boundaries.
  • 03
    Scope boundaries
    What is explicitly out of scope.
  • 04
    Acceptance signals
    How the organization knows the intent is fulfilled.
  • 05
    Supporting artifacts
    Design mocks, data tables, copy, diagrams.
§5.2 · Types of Intents

All intents flow through the same interface.

Whether the source is business, design, marketing, technical, or emergent, every intent is converted into the same format and fed into the Planner.

Business
Leadership · PM · Strategy

Launch a subscription tier · improve checkout conversion · enable international shipping

Design
UX · Brand

New visual hierarchy for onboarding · replace navigation · update accessibility patterns

Marketing
Growth · Revenue

Add referral incentives · seasonal campaigns · track key funnel events

Technical
Architects · Builders · Reviewers

Refactor a subsystem · address tech debt · update domain contracts · add observability

Emergent
PAT · Sev events · Reviewer feedback

Patch regression · fix semantic drift · replace failed Execution Plan

§5.4 · From Intent to Execution Plans

One intent, many atomic Execution Plans.

Intent sources
Business
Leadership · PM · Strategy
Design
UX · Brand
Marketing
Growth · Revenue
Technical
Architects · Builders · Reviewers
Emergent
PAT · Sev events · Reviewer feedback
Intent Queue
One per domain. Throttled by Planner capacity signals.
Planner
Interprets · sets boundaries · decomposes into atomic Execution Plans.
Execution Plans
Atomic
Testable
Deployable
Reversible
1 intent → many plans. Each ships independently through Ring 0 → Ring 1.
§5.5 · Atomicity

Every Execution Plan ships independently.

Atomicity is the foundation of continuous flow. A large intent may decompose into dozens of small Execution Plans, each shipped on its own. That keeps velocity high while preventing architectural collapse.

Atomic

Can be shipped independently.

Testable

Clear invariants and observable success signals.

Deployable

Moves through Ring 0 and Ring 1 on its own.

Reversible

If it fails, rolls back without entangling the broader intent.

§5.6 · Throttling

Intents flow downward, capacity signals flow upward. The Planner controls work entry to protect velocity.

No stakeholder, including PMs, can push work into the GI Flow past what the Planner authorizes.