What exactly is AiVIS Cite Ledger?

AiVIS Cite Ledger is an evidence-backed AI citation audit and remediation system. It measures how AI models, ChatGPT, Perplexity, Claude, and Google AI, interpret, trust, and cite your web content. The result is a 0–100 CITE LEDGER score across seven weighted dimensions, with BRAG evidence-linked findings and prioritized fixes.

How is this different from Semrush, Ahrefs, or Clearscope?

Traditional SEO tools optimize for keyword ranking in traditional search results. AiVIS Cite Ledger audits citation readiness, the specific signals that determine whether AI answer engines retrieve and cite your content. A page can rank #1 on Google and be completely invisible to AI answer synthesis.

What is BRAG and how does it make findings trustworthy?

BRAG stands for Based-Retrieval-Auditable-Grading. It ties every audit finding to a real, traceable element on your page, a specific heading, schema block, meta tag, or content section, each carrying a unique evidence ID. You can open the element, verify the finding, and recheck it after a fix. No black boxes.

Is the free tier genuinely useful, or just a teaser?

The Observer (free) plan gives you a full composite CITE LEDGER score and all seven dimension letter grades, so you can see exactly where you stand and which categories are hurting you. BRAG evidence IDs and prioritized fix packs require a paid plan. Three audits per month, no credit card.

See Exactly Where Your Brand Ranks in AI Answers, Query by Query, Model by Model | AiVIS Cite Ledger Blogs

By R. Mason · 2026-04-15 · 9 min read · IMPLEMENTATION

Stop guessing where your brand appears in ChatGPT, Claude, and Gemini. Run real queries, get real positions, and get a signed evidence trail for every result. Citation rank, not estimated, verified.

Key Takeaways

How the MCP validate_queries_evidence tool works end-to-end
What an evidence ID encodes and why it is deterministic
How brand detection handles numbered lists and prose responses
Position scoring: rank 1 = 100 points, decay by 5 per position
How to integrate citation rank data into agent workflows

Article

What the validate_queries_evidence MCP tool does

The `validate_queries_evidence` tool is available to Alignment and Signal tier users through the AiVIS Cite Ledger MCP server at `/api/mcp/call`. It accepts a brand name, a brand URL, and a set of test queries, then runs each query across live AI models and extracts structured evidence from every response.

The output is a CitationRankScore, a probabilistic 0-100 score representing your brand's AI citation strength, plus a full evidence record for every query × model combination tested.

This is not a simulated score. It is computed from actual model responses at the time of the call.

MCP call structure

```json

{

"name": "validate_queries_evidence",

"arguments": {

"brand": "Acme Corp",

"url": "https://acmecorp.com",

"queries": [

"best project management tools for remote teams",

"alternatives to Asana for small teams",

"how to manage sprints without JIRA",

"lightweight kanban tools 2026"

]

}

```

The tool accepts 1 to 20 queries per call. Each query is run against the AI models allocated to your tier:

**Alignment tier**: GPT-5 Nano (1 model per query)
**Signal tier**: GPT-5 Nano + Claude Haiku 4.5 + Gemini 2.5 Flash (3 models per query)

Signal tier coverage gives you cross-model validation, the ability to see whether a citation is consistent across model families or isolated to one.

How brand detection works in model responses

AI model responses take two forms: ranked lists and prose.

**Numbered list parsing**: The tool parses numbered list items using the pattern `/^\s*(\d+)[.)]\s+(.+)/`. When your brand appears in a numbered list, it records the exact position (1-indexed). Position 1 in a list of recommended tools is captured as `rank = 1`.

**Prose detection**: For responses where the brand appears in running text rather than a list, a position estimate of 10 is used, acknowledging the presence without assuming a specific competitive rank. This is a

Enable JavaScript for the full interactive reading experience with related articles and discussion.