Skip to main content
PLATFORM INTELLIGENCE

How RG Pulse Works

RG Pulse is an AI perception intelligence platform. It measures how frontier large language models collectively understand, describe, and recommend brands — and gives you the tools to shape that narrative. Here's exactly how the intelligence is generated.

6 LLMs37+ PromptsReal-time AnalysisHallucination DetectionNarrative Control

The Core Concept

AI Models Are the New Search

Millions of people now ask ChatGPT, Gemini, Claude, and Perplexity questions your brand used to answer via Google. What those models say about you shapes decisions — before a human ever visits your website.

Training Data Is the Battleground

LLMs form their views from web content, news, reviews, and documentation ingested during training. Gaps, errors, and competitor content in that corpus directly determine what AI says about your brand today.

You Can Influence What AI Knows

By identifying exactly what models believe (and misbelieve) about your brand, you can publish targeted content that corrects hallucinations, fills gaps, and shifts the AI narrative — measurably.

The 6-Step Intelligence Pipeline

We dispatch a battery of 37 standardised prompts across every topic dimension — from general perception and trust to innovation, ESG, competitive stance, and AI-specific framing. Prompts are templated with the target brand/subject and optionally enriched with Cognitive Profile system prompts to simulate specific audience lenses.

37 standard prompts across 10 categories
Custom prompts per theme or persona
Cognitive Profile system-prompt injection
Multi-language variants for global scans

Understanding Your Scores

Consensus Score0–100

The headline metric. Weighted average of sentiment scores across all 6 LLMs for a given scan. Above 66 = Strong, 41–66 = Mixed, below 41 = Weak.

Visibility Score0–100

How frequently the brand is mentioned — and mentioned positively — when relevant queries are posed. Low visibility = LLMs ignore the brand even when relevant.

Narrative Control Score0–100

How much the brand controls its own story. High = LLMs use the brand's own framing. Low = LLMs use competitor or third-party framing.

Divergence Score0–100

Standard deviation across model scores. Low divergence = consistent AI perception. High divergence = training data inconsistencies and hallucination risk.

Sub-Scores0–100 each

Dimensional breakdowns across: Financial Health, Trust & Reputation, Competitive Position, Innovation, Talent, Customer Experience, ESG, Growth Outlook, Media Narrative, AI Perception.

Platform Capabilities

Themes

Scope scans to specific topics — autofocus, sustainability, a product launch — and track sentiment evolution per theme over time.

Cognitive Profiles

Simulate how different audience types (wire photographers, brand managers, equipment financiers) perceive your brand through custom LLM personas.

Vulnerability Scoring

Adversarial probing to identify which narratives LLMs can be manipulated to believe about your brand, and what to do about it.

Global Scan

Run scans in multiple languages simultaneously to surface geo-specific perception gaps that only appear in local-language queries.

Prompt Gaps

Identify queries where competitors appear but your brand doesn't — the highest-leverage opportunities to recapture AI visibility.

Narrative Map

Visualise how themes cluster and relate across LLM responses — a network graph of your brand's AI narrative landscape.

Content Strategy

AI-generated content briefs tied directly to detected gaps and hallucinations, with suggested channels, keywords, and expected impact.

Narrative Repair

Generate and simulate the impact of counter-narratives before publishing — predict the lift in Consensus Score before committing.

The AI Model Panel

Every scan queries 6 frontier models simultaneously. Each model has different training data, different strengths, and different biases — which is precisely why using all of them together is more powerful than any single model alone.

GPT-5

OpenAI

Broadest web knowledge, strong on corporate narratives

GPT-5 Mini

OpenAI

Faster, cost-efficient — useful for high-volume prompt testing

Claude Sonnet 4.6

Anthropic

Nuanced reasoning, strong on tone and editorial framing

Gemini 3 Flash

Google

Google index-informed, strong on recent and technical content

Gemini 3.1 Pro

Google

Deeper reasoning, better at multi-step competitive analysis

DeepSeek V3

DeepSeek

Strong alternative training corpus — surfaces unique divergences

GET STARTED

Ready to scan your brand?

Run your first scan in under 2 minutes. Choose a company, pick a persona, and see exactly what AI models say about you.

Run a Scan