Project

Vocarum

Transparent consensus journalism that combines multiple sources and independent AI interpretations into traceable, versioned evidence reports.

Status: prototypeReviewed: 7/16/2026

Transparent consensus journalism

Understand where evidence converges—and where it does not.

Vocarum is an AI-assisted journalism and evidence-publication platform that compares how multiple independent models interpret a body of source material. It surfaces stable areas of agreement, preserves meaningful disagreement, and publishes the resulting analysis with source provenance and version history.

Readers should be able to inspect the evidence behind each conclusion rather than receiving a detached summary. Vocarum connects published analysis back to source documents, extracted claims, model interpretations, provenance metadata, and prior report versions.

The opportunity

Public evidence is increasingly difficult to audit.

Readers encounter fragmented reporting, incompatible narratives, incomplete context, and summaries whose assumptions are difficult to inspect. Vocarum approaches that challenge constructively: by making evidence collection, interpretation, agreement, disagreement, and publication history visible.

Fragmented evidence

Important facts are distributed across reports, statements, research, public records, and reporting.

Hidden assumptions

Different sources often frame the same event through different definitions, priorities, or contextual assumptions.

Unclear disagreement

Disagreement may be factual, definitional, methodological, temporal, or contextual rather than a simple contradiction.

Detached conclusions

Published conclusions can become separated from the evidence, scope, and analytical steps used to produce them.

The Vocarum approach

Turn source collections into reviewable evidence maps.

Vocarum is a transparent evidence-intelligence platform that helps readers understand where independent sources and analytical models agree, where they differ, and how published conclusions were produced. The goal is not to suppress disagreement, but to make agreement and disagreement visible, traceable, and interpretable.

Source aggregation

Collect documents from defined public or approved source sets for scoped analysis.

Multi-model interpretation

Use multiple independent AI models to interpret the same body of material.

Claim extraction

Identify claims, factual assertions, context, framing, and source-linked evidence.

Consensus mapping

Surface interpretations that remain stable across source material and analytical systems.

Divergence analysis

Preserve meaningful differences in facts, definitions, methods, context, timing, or interpretation.

Versioned publication

Publish reports with source references, provenance, timestamps, version identifiers, and lineage.

How it works

A structured path from source collection to published report.

  1. Collect documents from defined public or approved sources through scoped collection or scheduled ingestion.
  2. Extract claims, factual assertions, context, and framing from each source.
  3. Ask multiple independent AI models to interpret the same material.
  4. Compare outputs to identify stable overlap and meaningful divergence.
  5. Classify disagreement by type, such as factual, definitional, methodological, contextual, or temporal.
  6. Publish a traceable report with source references, model-output provenance, timestamps, and version history.

Product experience

Consensus should be inspectable, not merely asserted.

A reader should be able to move from a published conclusion back to the source documents, extracted claims, model interpretations, and prior versions that contributed to it. Vocarum is designed to keep the analytical trail available for readers, researchers, editors, and institutions.

Source-linked claims

Claims remain connected to the documents and passages that informed them.

Areas of model agreement

Stable overlap across independent interpretations is highlighted as an analytical signal.

Disagreement zones

Readers can inspect where interpretations differ and how those differences are classified.

Alternative interpretations

Competing readings of the same material are preserved when they remain meaningful.

Publication history

Report versions, timestamps, and update history disclose how analysis changes over time.

Version diffs

Differences between report versions can show what was added, revised, or superseded.

Provenance metadata

Evidence, source-set, and model-output metadata help explain how a report was constructed.

Scope limitations

Corpus boundaries and analytical limits remain explicit so conclusions are not overextended.

Consensus and divergence

Agreement is a signal. Disagreement is information.

Multi-model agreement can identify interpretations that remain stable across different analytical systems, but model agreement does not guarantee truth. Vocarum treats consensus as a useful signal to inspect—not as a substitute for evidence, editorial judgment, or expert review.

Divergence can reveal ambiguity, missing evidence, different definitions, temporal differences, or competing interpretations. Vocarum preserves both overlap and disagreement so readers can evaluate the evidence themselves, while recognizing that human editorial or expert review remains important for high-stakes publication.

Tamper-evident publication

Published reports should disclose the record that produced them.

Vocarum's publication architecture is oriented around manifests that make later changes visible and allow a reader or publisher to verify which source set and analytical outputs produced a particular report version.

  • Signed publication manifests: Record the declared report package and publication metadata when signing is enabled.
  • Content hashes: Identify source bundles, analytical outputs, and report artifacts without describing hashes as encryption.
  • Timestamps and version identifiers: Distinguish one report version from another.
  • Source references: Preserve the source set used to generate the analysis.
  • Report lineage: Show which prior versions, updates, and analytical outputs contributed to publication.
  • Optional redundant publication: Support future or optional publication paths through systems such as IPFS where appropriate.

Applications

Evidence intelligence for public-interest knowledge work.

Independent journalism

Create traceable reporting that preserves both evidence overlap and meaningful disagreement.

Policy analysis

Compare official statements, reports, legislation, outcomes, and later revisions.

Science communication

Translate complex research while retaining source provenance, methodological limits, and conflicting findings.

Education

Produce reviewable learning materials that distinguish broadly supported knowledge from unresolved questions.

Institutional research

Create auditable evidence reports from approved public or private corpora.

Public-interest archives

Preserve evolving narratives, supporting documents, and version history over time.

Current implementation

An operational MVP with a clearly scoped roadmap.

Vocarum currently has an MVP ingestion-to-publication pipeline that supports document collection, model interpretation, intersection analysis, and publication. Contradiction and divergence classification is in beta, with classification quality and review workflows continuing to mature.

Manifest-based publication is part of the active architecture direction, including source references, report lineage, timestamps, version identifiers, and content hashes. Redundant decentralized storage, including optional IPFS publication, remains optional or planned depending on deployment needs. Peer-to-peer model evaluation through WhispersNet is a future integration direction and is not presented here as a current integration.

Research record

Protected architecture and published research.

Vocarum is supported by both a published research record and an active intellectual-property strategy. Its public research or archival record is hosted through Zenodo at https://doi.org/10.5281/zenodo.16636107.

Its underlying architecture is covered by U.S. Provisional Patent Application No. 63/873,534, addressing ontology-framed multi-model consensus reporting, model disagreement analysis, and verifiable publication manifests. Procyonsoft is preparing the corresponding non-provisional patent application as the platform advances.

Long-term vision

A public evidence layer for an increasingly complex information world.

The long-term vision for Vocarum is an open and inspectable system for understanding how claims emerge, which evidence supports them, where independent analyses converge, and how interpretations change over time.

Vocarum aims to make published knowledge more transparent by connecting conclusions to sources, preserving disagreement, and giving readers a visible record of how each report was constructed. It is infrastructure for journalists, researchers, educators, institutions, and readers—not a replacement for all human journalism or expert judgment.

Start a conversation

Help build a more transparent evidence ecosystem.

Procyonsoft is opening conversations with journalists, researchers, public-interest organizations, educational institutions, investors, and strategic partners interested in traceable AI-assisted publishing.