The Engine

The intelligence architecture
behind PharosGraph.

Pharos combines AI, computational linguistics, behavioral science, narrative theory, and proprietary intelligence architecture to transform public discourse into decision-ready strategic intelligence.

  • Architecture 5 layers
  • Individuals scored 200M+
  • Neighborhoods 240,000
  • Refresh Every week
  • Data Multi-Source Intelligence
  • Language AI / LLM Interpretation
  • Behavioral Moral Foundations · Persuasion
  • Geographic Block-Level Resolution
  • Activation Commercial Outputs

The Limits of Conventional Intelligence

Conventional intelligence tools were built for an earlier era.

They were designed to count what already happened. They cannot explain why narratives are moving, where movement is happening, how messages land differently across audiences, or what action to take next.

Sentiment Snapshots

A snapshot in time

Surfaces what people said when asked.

Misses

Why opinion is moving. What's coming next. How sentiment varies block to block.

Social Listening

Mentions and volume

Tracks who's talking and how loudly.

Misses

Whether the volume signals real movement. What it means for the people you need to reach.

Dashboards

Activity visibility

Aggregates what's already been measured.

Misses

Strategic recommendations. What action to take, where, with which audience, and why.

Prediction Markets

Outcome pricing

Aggregates speculation about end-state outcomes.

Misses

The structural drivers of those outcomes. The narratives moving voters before sentiment shifts.

Strategic intelligence about narratives, not activity reports about mentions.

  • Why narratives are gaining or losing ground, not just that they exist.
  • Where movement is happening, at the resolution of a city block.
  • How issues land differently across audiences, geographies, and moral frames.
  • What strategic action to take next, calibrated to current narrative state.

Architecture

A layered intelligence system, built for live decision environments.

The Pharos engine combines five tightly integrated layers. Each layer adds a distinct dimension of intelligence. Together, they convert raw public discourse into decision-ready strategic intelligence. Signals move through them in order: ingest, interpret, score, segment, recommend, and measure, with the Activation layer carrying the last two.

01

Ingest

Data Layer

Multi-Source Intelligence Inputs

News, public discourse, legislative and election data, policy signals, paid and earned media, and custom client sources, all ingested live and modeled together.

NEWS · LIVE DISCOURSE · LIVE LEGISLATIVE · NATIONWIDE PAID + EARNED MEDIA

02

Interpret

Language Layer

AI & LLM Interpretation

Large language models trained for political and policy discourse interpret what's being said, how it's framed, and what it means in narrative context.

"public safety is getting worse..." FRAME · FEAR "affordability crisis" FRAME · STRAIN "opportunity to lead..." FRAME · HOPE → NARRATIVE STRUCTURE EXTRACTED

03

Score

Behavioral Layer

Behavioral Science & Moral Foundations

Decades of behavioral science put to work: why messages resonate, where persuasion is asymmetric, and how different audiences read the same words differently.

MORAL FOUNDATIONS · ISSUE WEIGHTING · ILLUSTRATIVE CARE .87 FAIRNESS .69 LOYALTY .48 AUTHORITY .32

04

Segment

Geographic Layer

Hyperlocal Resolution

Narratives don't move uniformly. Pharos scores narrative dynamics at the resolution of a Census Block Group, then rolls up to ZIP, district, DMA, and state.

240,000 NEIGHBORHOODS · SCORED WEEKLY CBG → ZIP → DMA → DISTRICT → STATE

05

Recommend + Measure

Activation Layer

Commercial Outputs

The stack converges into outputs teams can deploy: activation audiences, strategic briefings, geographic targeting, and recommendations that connect directly to media, outreach, and operations.

BRIEFING delivery AUDIENCES paid · field TARGETING geographic RECOMMENDATIONS where to lean in MEASUREMENT Narrative Alpha™

Science

Built on rigorous science, designed for commercial environments.

The Pharos engine is grounded in disciplines that have shaped how we understand language, persuasion, behavior, and population dynamics. Operationalized for real decision environments, not academic ones.

Discipline 01

Computational Linguistics

Modeling how language carries meaning, frame, and intent at scale across millions of signals.

At Pharos: Powers narrative structure recognition, framing detection, and theme extraction.

Discipline 02

Behavioral Science

Decades of research on how messages persuade, how audiences interpret, and where decision-making bends.

At Pharos: Drives audience interpretation models and persuasion asymmetry scoring.

Discipline 03

Narrative Theory

The structural patterns of how stories form, propagate, and reshape collective interpretation.

At Pharos: Frames how narrative momentum is detected, tracked, and projected forward.

Discipline 04

Moral Foundations Theory

A research-backed framework for the moral dimensions (care, fairness, loyalty, authority, sanctity) that shape how messages land.

At Pharos: Powers the Behavioral Layer's audience-resonance scoring.

Discipline 05

Population Modeling

Statistical projection from sampled signals to population-level dynamics, calibrated against ground-truth data.

At Pharos: Translates signal patterns to neighborhood-level intelligence across 240K block groups.

Discipline 06

Explainability & Audit

Engineering practices that make AI-driven intelligence traceable, defensible, and scrutable.

At Pharos: Every recommendation surfaces the underlying signals that drove it.

Modules

Modular intelligence, calibrated to the environment.

The engine surfaces through proprietary intelligence modules called Scapes. Each Scape applies the full architecture to a distinct decision environment, with vertical-specific data, models, and outputs.

Live

RaceScape™

Intelligence for live races.

Statewide, Federal, Mayoral, and Local race intelligence. Narrative momentum, geographic opportunity, persuadable audience modeling, and activation guidance.

Live

IssueScape™

Live tracking of issue dynamics.

Continuous monitoring of how issue narratives form, move, and land across geographies and audiences. Built for advocacy, donors, and policy teams.

Live

GovScape™

Strategic policy intelligence.

Translates legislative activity into constituent pressure, district movement, legislator alignment, and strategic outreach guidance. Sits above traditional bill-tracking tools.

Live

BallotScape™

Ballot measure intelligence.

Calibrated for ballot initiatives and referenda. Geographic activation modeling, narrative pressure detection, and measurement of campaign deployment performance.

Live

VoterScape™

The living voter file.

The enrichment layer for the file you already have. Every record comes back with a ranked issue list, a momentum signal with direction, and one message to lead with. Refreshed every week.

Live

AudienceScape™

Activation-ready audience segments.

Voter-level audiences built from narrative intelligence, split into GOTV and Persuade clusters with the lead issue for each neighborhood. Delivered through the ad-tech stacks your team already runs.

See the engine on your terrain

See Pharos in action.

Bring us a race, an issue, or an initiative. We will walk the architecture end to end on your terrain, from the signals that feed it to the recommendation that comes out.

More in How It Works: Narrative Command Center and Intelligence Briefs.