Six diagnostic lenses that explain why AI names, omits or misdescribes your brand.
Intelligence is the diagnostic layer. It holds the reviewed facts about your brand and the topics and prompts you track, then explains gaps through six lenses: visibility, semantic positioning, omissions, co-mentions, sources and factuality, plus regional and model variance. Four ship at launch and the rest deepen in V2.
Intelligence
Why is it happening?
- Brand Context
- Topics and Prompts
- Visibility Diagnosis
- Omissions
- Co-Mentions
- Source Intelligence
- Semantic Positioning
- Factuality and Risk
- Regional and Model Variance
Quick answer: Intelligence
Intelligence is the diagnostic layer of BrandWater AI. It holds the brand facts you confirmed and the topics and prompts you track, then explains gaps through six lenses: visibility, semantic positioning, omissions, co-mentions, sources and factuality, plus regional and model variance. Four ship at launch. Positioning, factuality and regional variance are coming in V2.
Six diagnostic lenses
Each lens asks one question of the same stored evidence. Four ship at launch. The rest deepen in V2.
- 1Available at launch
Visibility and share of voice
Does AI recognise and recommend the brand?
- 2Coming in V2
Semantic positioning
What concepts does AI associate with the brand?
- 3Available at launch
Omission and saliency
Which known strengths are rarely mentioned?
- 4Available at launch
Co-mention network
Which brands keep appearing with us?
- 5Coming in V2
Risk and factuality
Which recurring facts are wrong or outdated?
- 6Coming in V2
Regional and model variance
How do answers differ by market, language and engine?
What Intelligence gives you
9 capabilities, each answering a specific question and labelled with its release stage.
How Intelligence measures things
Every metric has a definition, so a number always means the same thing.
| Metric | Definition | Capability |
|---|---|---|
| Fact state | Discovered, confirmed or edited. | Brand Context |
| Evidence link | Each fact points to the source page it came from. | Brand Context |
| Coverage | How many of the expected brand attributes have an evidence backed value. | Brand Context |
| Topic set | The confirmed topics, each with rationale. | Topics and Prompts |
| Intent mix | The share of prompts per intent, so no single type dominates. | Topics and Prompts |
| Active prompts | The prompts that will run, and how many runs that means across engines. | Topics and Prompts |
| Visibility gap | The difference between observed and expected visibility for a slice. | Visibility Diagnosis |
| Scope | Prompt set, topic, intent, engine, market, language and time. | Visibility Diagnosis |
| Evidence | The run IDs, answers and sources behind the finding. | Visibility Diagnosis |
| Observed mention rate | How often an attribute is mentioned in relevant answers. | Omissions |
| Expected saliency | High, medium or low, from the importance of the attribute to the business. | Omissions |
| Gap status | Flagged when observed saliency is materially below expectation. | Omissions |
| Co-mention frequency | Runs that contain both brand A and brand B. | Co-Mentions |
| Conditional co-mention | Runs with both, divided by runs with A, in the same scope. | Co-Mentions |
| Alternative rate | Share of prompts asking for alternatives to A where B appears. | Co-Mentions |
| Citation rate | Eligible runs citing your domain divided by eligible runs. | Source Intelligence |
| Source gap | A topic where competitors are cited and your domain is not. | Source Intelligence |
| Source class | The type of source, such as review, publisher or directory. | Source Intelligence |
| Concept frequency | How often a concept appears in answers about a brand. | Semantic Positioning |
| Cluster share | The share of answers in each theme. | Semantic Positioning |
| Association gap | The difference between your association and a competitor's for a concept. | Semantic Positioning |
| Claim status | One of six verification statuses. | Factuality and Risk |
| Recurrence | How often a claim appears across runs. | Factuality and Risk |
| Confidence | How certain the classification is, with evidence. | Factuality and Risk |
| Variance | Spread of a metric across repeat runs or segments. | Regional and Model Variance |
| Consistency | How often repeat runs agree. | Regional and Model Variance |
| Completeness | The share of intended runs that completed. | Regional and Model Variance |
How Intelligence connects
When a diagnosis points to a fix, it becomes an opportunity in Agent.
Questions it answers
- What does the platform know about us, and is it right?
- Where and why does our visibility break?
- Which of our strengths does AI leave out?
- Which brands and sources keep appearing with us?
What it is not
- Executing changes
- Intelligence explains. Agent recommends, and from V3 executes with approval.
- Reading a model's internals
- It analyses observed answers and cited sources. It never claims access to a model's private data or reasoning.
- Replacing your judgment
- Every finding shows its scope and evidence so a person can accept or reject it.
Intelligence questions
Which lenses are available at launch?
Visibility and share of voice, omissions, co-mentions and source intelligence, plus the brand context and topic and prompt setup they depend on.
How does it explain why?
By comparing observed answers with cited sources and your confirmed brand record. Interpretation is labelled as interpretation.
Can we correct a fact?
Yes. Facts are discovered, then confirmed or edited by you, and your edits are never silently overwritten.
See Intelligence on your own brand.
Start with a first audit. We show where you appear, who is named instead, and what to fix first.