Guide2 min read

How Agents Track Brand Mentions in AI Search

Track how ChatGPT, Claude, Gemini, and Perplexity mention your brand with AgentAEO — LLM responses, mention search, and AI keyword volume per tenant.

Read the docs →

Guide
TL;DR
  • Buyers now ask ChatGPT and Perplexity before they Google brand teams need to know whether AI answer engines mention and cite them
  • AgentAEO (/aeo) exposes llmResponses across ChatGPT, Claude, Gemini, and Perplexity plus llmMentions search scoped to your domain
  • Run a tracking loop per brand tenant: prompt set → LLM responses → mention share → alert when competitors gain citations
  • Scope with naive.forUser(brandId) so one customer's audit never leaks into another's workspace
  • Pair with AgentSEO keyword loops for classic SERP data AEO answers what models say, SEO answers what people type
  • Meter aeo calls per plan when you productize same Customer Billing pattern as the usage-metered SEO SaaS tutorial

When a buyer asks ChatGPT "what's the best project management tool for agencies?", the model returns an answer — and a short list of cited sources. That response is the new SERP for a growing share of research. Brand teams need to track whether they are mentioned and cited, not just where they rank on Google.

Naïve ships AgentAEO (/aeo) so an agent can run that audit on a schedule: query multiple answer engines, measure mention share, and alert when a competitor gains ground — per brand tenant, not from a shared scraper account. This use case sits in the autonomous company spoke cluster alongside SEO keyword loops and the full AEO platform tutorial.

Brand mention tracking loop per tenant on Naïve

Tracking loop

StepAgentAEO callPurpose
1. Prompt setYour app stores category questionsStable audit baseline
2. LLM responsesllmResponses("chatgpt" | "claude" | …)Capture answers + citations
3. Mention searchllmMentions("search", { keyword, domain })Quantify your domain's share
4. AI keywordsaiKeywords({ keywords })Rising AI-search demand
5. ReportEmail, dashboard, or kanban taskHuman-readable delta

Run weekly or after major launches. Store snapshots so the agent can diff last week vs this week.

Setup

npm install @usenaive-sdk/server
import { Naive } from "@usenaive-sdk/server";
 
const naive = new Naive({ apiKey: process.env.NAIVE_API_KEY! });
 
const brand = await naive.users.create({
  external_id: "northwind_uuid",
  email: "marketing@northwind.example",
  label: "Northwind",
});

Capture answers across engines

const PROMPTS = [
  "best project management software for agencies 2026",
  "top CRM for small business",
];
 
async function auditAiVisibility(brandId: string, domain: string) {
  const aeo = naive.forUser(brandId).aeo;
  const snapshots = [];
 
  for (const user_prompt of PROMPTS) {
    const chatgpt = await aeo.llmResponses("chatgpt", {
      user_prompt,
      model_name: "gpt-4o-mini",
    });
    const mentions = await aeo.llmMentions("search", {
      keyword: user_prompt,
      domain,
    });
    snapshots.push({ user_prompt, chatgpt, mentions });
  }
 
  return snapshots;
}
  • CLI mirror: naive aeo llm-responses chatgpt --keyword "<prompt>" and naive aeo llm-mentions search --keyword "<prompt>" --domain example.com.
  • Full parameter list: AEO docs.

Combine with SEO research

A complete visibility agent runs both loops:

  1. AgentSEO keyword research — volume + SERP competitors.
  2. AgentAEO mention tracking — AI answers + citation share.

The multi-tenant AEO platform tutorial wires both into one product with Customer Billing quotas. For metering details see usage-metered SEO SaaS.

Productizing the loop

When you sell this as a service:

  • One tenant user per brand — hard isolation boundary.
  • Account Kits — gate which platforms a tier may audit.
  • Customer Billing — meter aeo calls; return 429 rate_limited at quota.

Sensitive actions (publishing revised copy to a live CMS) should stay approval-gated even when the audit itself runs autonomously.

Related reading

Frequently Asked Questions
What is brand mention tracking in AI search?+
It is monitoring whether AI answer engines name or cite your brand when users ask category questions — for example 'best CRM for agencies'. Unlike classic rank tracking, the output is a natural-language answer plus cited URLs. Naïve's /aeo primitive returns those answers programmatically and supports mention search filtered by domain.
Which AgentAEO calls power a tracking loop?+
Use aeo.llmResponses(platform, { user_prompt, model_name }) to capture what each engine says today. Use aeo.llmMentions or the CLI naive aeo llm-mentions search --keyword ... --domain yourbrand.com to see mention share. AI keyword volume via aeo.aiKeywords shows rising AI-search demand for a topic. See usenaive.ai/docs/getting-started/aeo for costs and parameters.
How is this different from SEO keyword research?+
SEO tracks ranked pages on Google and Bing; AEO tracks answers inside LLMs. Many teams run both loops: AgentSEO for volume and SERP competitors, AgentAEO for AI visibility. Naïve exposes both primitives on the same tenant user so one agent can run a combined brief.
How do I isolate one brand from another?+
Create one tenant user per brand with naive.users.create. Every aeo and seo call goes through naive.forUser(brandId). Account Kits gate which engines or actions a tier may use. There is no shared API key in the agent prompt.
How do I get started?+
Install @usenaive-sdk/server, define a prompt set for your category, and run the loop below with naive.forUser(brandId).aeo. Docs: usenaive.ai/docs/getting-started/aeo.
NT
Naïve Team

Building the autonomous company infrastructure.

Keep reading