The methodology

How we govern the AI narrative.

A five-step method to see what AI says about you, measure where it’s wrong, and put it right — on loop.

01 · Build Corporate Truth

02 · Create Querying Personas

03 · Interrogate multiple LLMs

04 · Analyze discrepancies

05 · Generate insights & actions

01 / 05 · Foundation

Build Corporate Truth

✓ Structured, attributable and version-controlled.

✓ Reviewed and signed off by your own teams.

✓ The benchmark AI is measured against.

Corporate Truth · recordVerified

Founded2014 · Milan, ITSigned

LeadershipA. Reyes, Chief ExecutiveSigned

PositionEthical materials, full traceabilitySigned

Footprint11 markets · 1,400 employeesSigned

02 / 05 · Perspective

Create Querying Personas

✓ Investors, candidates, customers, journalists, regulators.

✓ Hundreds of realistic prompts per persona.

✓ Mapped to the decisions each stakeholder makes.

Querying Personas4 active

The InvestorDue diligence“Any governance red flags?”

The CandidateEmployer brand“What's it like to work here?”

The CustomerPurchase intent“Are they reliable?”

The JournalistScrutiny“Any past controversies?”

03 / 05 · Interrogation

Interrogate multiple LLMs

✓ Coverage across every leading generative system.

✓ Repeated over time to catch drift.

✓ Every answer captured, attributed and timestamped.

Multi-model interrogationlive

“Is this company a responsible employer?”

ChatGPT“Generally regarded as progressive…”

Gemini“Limited public information…”

Claude“Strong sustainability record…”

Perplexity“Faced scrutiny over suppliers…”

04 / 05 · Analysis

Analyze narrative discrepancies

✓ Consistency, accuracy and sentiment scored per model.

✓ Hallucinations isolated and evidenced.

✓ Drift visualized over weeks and months.

Discrepancy analysisvs. corporate truth

Factual accuracy68%

Cross-model consistency54%

Sentiment alignment81%

Hallucination rate12%

05 / 05 · Action

Generate insights & actions

✓ Prioritized by impact and effort.

✓ Owned, tracked and measured to completion.

✓ Re-verified against the models after intervention.

Recommended actionsprioritized

HIGH

Correct leadership records across modelsCloses a hallucinated-executive gap surfaced in 3 of 4 models.

HIGH

Publish an authoritative supplier statementCounters the negative drift detected over the last 6 weeks.

MED

Clarify product claim languageResolves inconsistent descriptions between Gemini and Claude.

A volume no team can handle

A network of AI agents works for you. You just see the results.

DICTO isn’t a passive dashboard. It’s an agentic system: specialized agents that split the work, coordinate, and hand back only what needs a decision.

QUERYING AGENTS

Probe the models from dozens of angles, continuously.

ANALYSIS AGENTS

Extract claims, sources and framing; compare them to your truth.

WATCH AGENTS

Watch critical topics and fire alerts when something changes.

RECOMMENDATION AGENTS

Turn findings into ready-to-assign actions.

AUTHORING AGENTS

Generate correction assets, ready for models to ingest.

Closed-loop governance

It never stops at a single report.

Models change every week, so we run a continuous loop — six stages that turn each analysis into verified change, then start again.

01 Analysis
Continuous interrogation of every model across all personas.

02 Diagnosis
Discrepancies, hallucinations and drift classified and quantified.

03 Recommendations
Prioritized interventions tied to the gaps they close.

04 Intervention
Coordinated action across your owned and earned channels.

05 Publishing
Authoritative, structured truth released for models to ingest.

06 Verification
A fresh interrogation confirms the narrative moved — and the loop restarts.

See the methodology run on your own organization.

Request a confidential briefing and we’ll walk your team through every step on your own narrative.

Reputation

Go-to-market

Organization