Methodology

How PIGI
works.

PIGI (Pattern Intelligence Graph Interface) is an analytical system designed to identify and expose relationships, patterns and structural differences across supplied information. PIGI is a mirror, not a judge.

Version 1.0 · Last updated 2026-09-28 · Canonical source: PIGI.app

01

What PIGI does

Its purpose is not primarily to generate opinions or verdicts. Its purpose is to make relationships in information easier to inspect.

Its canonical question: Does what you say match what you do? — the relationship between declarations, execution, framing and observable value flows, not just the detection of contradictions.

02

Why relationships matter

Individual facts are useful. Relationships between them reveal structure.

PIGI does not only ask what information exists. PIGI asks how the information relates.

03

The four lenses

Declaration
What is said, claimed, promised, stated, described or presented.
Execution
What actions, outputs, behaviors or observable events appear in the supplied material.
Framing
How information is presented, contextualized, emphasized, omitted, positioned or linguistically shaped.
Value Flow
Where observable or proposed forms of value appear to move — money, attention, information, access, authority, reputation, opportunity, time, resources, influence or other context-dependent value. Value flow does not automatically establish motive.

04

From material to structure

  1. Material
  2. Observable information
  3. Information units
  4. Entities + Signals
  5. Relationships
  6. Patterns
  7. Questions / Exploration

This is the conceptual model, not the private engine. Source code, prompts and internal architecture are not published.

05

Units, entities, signals, relations, patterns

Information Unit
A smaller piece of material — a statement, observation, claim, event, description, action or relationship-bearing fragment. Units let PIGI reason about structure rather than merely summarize whole documents.
Entity
A distinguishable object in the material: a person, organization, company, product, document, publication, project, role, system or event. Same label does not necessarily mean same entity.
Signal
A relevant informational feature detected or supplied within the material. A signal is not a conclusion; it becomes useful when connected to its source, scope, entity and other signals.
Relation
How supported pieces of information connect — for example supports, contradicts, precedes, follows, references, repeats, changes, benefits or depends on. PIGI does not only ask what information exists; it asks how the information relates.
Pattern
A structure emerging across relations — repetition, change over time, consistency, inconsistency, asymmetry, missing expected information, gaps between declaration and execution. A pattern is not a verdict; it is a structure worth examining.
Unknown
An explicit state meaning the material does not support an answer. Unknown must remain possible; missing information is not permission to invent.

06

Value flows

Where observable or proposed forms of value appear to move — money, attention, information, access, authority, reputation, opportunity, time, resources, influence or other context-dependent value. Value flow does not automatically establish motive.

Benefit does not establish intention. A possible value flow is not an observed value flow.

07

Observation vs interpretation

Observation
What appears directly in supplied material.
Relation
How supported pieces of information connect.
Pattern
A structure emerging across relations.
Interpretation
A possible meaning assigned to that structure.
Conclusion
A stronger claim that requires sufficient support.

PIGI does not forbid conclusions. PIGI forbids invisible transitions.

08

Epistemic boundaries

These are part of the analytical methodology, not marketing disclaimers.

  • Unknown must remain possible.
  • Missing information is not permission to invent.
  • Correlation ≠ causation.
  • Benefit ≠ motive.
  • Function ≠ intention.
  • Repetition ≠ independent confirmation.
  • Same label ≠ same entity.
  • Same label ≠ same signal.
  • Possible value flow ≠ observed value flow.
  • Relationship ≠ permission for any conclusion.

09

One source vs many

One piece
Observations within that material.
More pieces
Relationships between them.
Over time
Patterns and trajectories.

More material does not imply certainty.

10

Human judgment

The human remains responsible for context, judgment, verification, decisions and appropriate use of information.

PIGI makes relationships difficult to unsee.

11

What PIGI does not claim

PIGI is not primarily a generic chatbot, a search engine, a fact checker, a scoring engine, a social-media monitoring platform, a personality test, a surveillance system, a background-check database, a simple summarizer, an automatic truth machine, an automatic lie detector.

Canonical category: relationship and pattern analysis system / interface. PIGI can be wrong; verify important facts against primary sources.

12

Example

A company statement declares a commitment to transparency (Declaration). Its published reports show which actions occurred (Execution). The language emphasizes some facts and omits others (Framing). Attention and reputation move toward certain parties (Value Flow).

PIGI surfaces how these relate and which questions follow. It does not declare the company honest or dishonest.