Domain Pattern Intelligence (DPI)

A structured methodology that teaches and applies the discipline of producing AI-augmented research that is valid, defensible, and authoritative within a specific domain. Built on a three-component core framework and operationalised through a scored instrument set.

Authority Mapping

Identifying the hierarchy of validating institutions, standards bodies, regulatory frameworks, and cultural arbiters that determine whether a finding is credible in a given domain.

Contextual Pattern Recognition

Encoding the domain-specific patterns, speech acts, temporal contexts, cultural variables, and structural signals that trained practitioners recognise and that AI systems miss without explicit instruction.

Validation Chain Construction

Building a structured, repeatable sequence of checks that every AI-generated output must pass before it is considered credible, citable, or actionable.

Same core framework. Different ways to apply it.

DPI isn't one workshop product - it's the same three-component discipline instantiated four ways, depending on whether you want to learn it, buy it pre-calibrated, have it applied on your behalf, or embed it as a standing layer over your own AI use.

Core Methodology Training

The three-component framework and its supporting instruments, taught domain-agnostically.

  • Half-Day Workshop
  • Full-Day Intensive
  • Two-Day Organisational Program
  • Virtual Workshop Series
  • Self-Paced Online Course
  • Licensing to training providers

Industry / Company Defaults

Pre-calibrated domain instantiations - fixed authority maps, pattern libraries, and instrument weights for a named vertical, or bespoke-calibrated to your internal governance standard.

  • Social Research & Human Rights (in development)
  • Financial Services
  • Legal Research
  • Policy & Government
  • Corporate Intelligence

Context-Based Analysis

The SENTIUS Research Suite viewed through the DPI lens - DPI as the internal validation discipline behind a delivered analytical product, rather than a separately sold methodology.

DPI Assure

An embedded validation and compliance layer over your own AI pipeline, mapped against named external standards. Jump to DPI Assure.

  • Validation Design Engagement
  • Validation Retainer
The instrument set

Ten scored instruments, each testing a different way a conclusion can fail.

DPI doesn't ask a model to hold an entire methodology in its attention at once. Each analytical stage is a discrete, reviewable gate - the output of one gate constrains what the next gate is permitted to assert.

RODI

Recorded–Observable Divergence

Scores the gap between what's claimed or reported and what the observable record actually shows.

OIV-S

Identity Divergence

Tests a declared or recorded identity against observable characterisation across six standard dimensions.

AFM-F

Formation Closure Index

Scores how open or closed an actor's formation environment is to correction under contradicting evidence.

JAT-L

Causal Inversion Likelihood

Tests whether a stated justification's cause-and-effect sequence is upright, indeterminate, or inverted.

CWC

Corroboration & Confidence

Aggregates independent evidential streams into a confidence level - repetition from one source chain doesn't count, regardless of volume.

CACS

Compounded Adverse Score

Detects when multiple instruments simultaneously produce adverse findings and escalates the evidentiary bar accordingly.

ACD-S

Authority-Capacity Divergence

Scores the gap between an actor's declared authority and its actual, observable operational capacity.

QTD

Query-Target Divergence

Catches mismatched or drifting terminology before it's allowed to quietly shift what's being analysed mid-session.

SPV

Source Provenance Validation

Tests whether a source is genuinely independent or only presents that way - the specific gap a coordinated content campaign is built to exploit.

ICC

Institutional Capacity & Coordination

Tests claimed institutional incapacity or coordination against an actor's demonstrated behaviour elsewhere, rather than assuming either by default.

Formulas, weight-declaration methodology, and full gate templates are part of the applied methodology and are shared under briefing or engagement - this page describes what each instrument tests, not how it's scored.