How the 5 Agents Work

Genezio runs five types of AI conversations to understand your brand's presence. Each type answers a different question — and not all of them count toward your score. Here's why that's a feature, not a limitation.

Think of it like this: Imagine hiring five different mystery shoppers to evaluate your brand. Each one walks into a store (or asks an AI) with a different brief. The results they bring back reveal very different things about how you're perceived.

Agent 1: Prompter Agent

Counts toward: AI Visibility %

What it simulates: A user at the top of the funnel — exploring a category, not yet committed to any brand. They ask open-ended discovery questions.

Example queries:

• "What are the best CRM tools for a startup?"
• "What software should I use for project management?"
• "What tools do marketing teams use for analytics?"

Why it matters for you: This is how most buyers start their research. If you're absent here, you're being filtered out before the shortlist even forms. Prompter conversations are the biggest driver of AI Visibility %.

Content implication: Category pillar pages · "Best of" content · In-depth guides · Original research

Agent 2: Recommender Agent

Counts toward: AI Recommendations % and AI Visibility %

What it simulates: A user ready to make a decision. They've done some research and now they want the AI to pick something specific for their situation. These are often multi-step conversations.

Example queries:

• "I'm a solo founder with a team of 3. Recommend a CRM that's easy to set up and under $50/month."
• "What's the best project management tool for a remote design team of 10?"

Why it matters for you: This is the highest-intent moment in the AI buyer journey. If a user asks AI to recommend something and your name doesn't come up, you've lost a qualified lead. AI Recommendations % is your conversion metric.

Content implication: Use-case landing pages · Persona-specific content · ROI calculators

Agent 3: Introspector Agent

Does not count toward KPIs

What it simulates: A user who already knows your brand and wants to learn more. They ask the AI directly about you by name.

Example queries:

• "What is HubSpot?"
• "What does Salesforce do?"
• "Tell me about Notion's pricing model."

Why it doesn't count toward your score: If the user mentions your brand in the question, the AI will almost always mention it in the answer — that's not meaningful. Including these would artificially inflate your Visibility score and make it useless as a benchmark.

What it's useful for instead: Understanding how AI describes your brand. What claims does it make? Are those claims accurate? This feeds directly into your Perceptions analysis — the accuracy layer underneath the score.

Content implication: Brand narrative analysis · Claim accuracy monitoring · Positioning audit

Agent 4: Comparer Agent

Does not count toward KPIs — this is a competitive analysis tool

What it simulates: A user actively evaluating you against a specific competitor. They've narrowed their options and want a head-to-head verdict.

Example queries:

• "HubSpot vs Pipedrive: which is better for a startup?"
• "Notion vs Asana for a remote team?"
• "Should I use Salesforce or HubSpot?"

Why it doesn't count toward your score: Both brands are named in the question, so both will always appear in the answer. Like the Introspector, including these would distort your metrics.

What it's actually for: The Comparer is an analysis tool. It takes the brand-level KPIs that Genezio has already calculated from Prompter and Recommender conversations and puts them in competitive context. It also runs head-to-head AI conversations to reveal how answer engines frame you against specific competitors — what strengths it highlights, what weaknesses it mentions, and who it picks as the better fit.

Think of it this way: Prompter and Recommender tell you your score. Comparer tells you why you're winning or losing against a specific rival.

Content implication: Competitive intelligence · "vs" comparison page strategy · Win/loss narrative

Agent 5: Fact Checker Agent

Does not count toward KPIs — this is an accuracy measurement tool

What it simulates: A factual probe. Instead of asking the answer engine open-ended questions, you give it specific verifiable claims about your brand (pricing, features, founding year, customer counts, certifications) and measure whether it confirms, contradicts, or dodges each one.

Example claims:

• "Our product offers a free tier."
• "We were founded in 2019."
• "We are SOC 2 Type II certified."

Per-claim outcomes:

• True — the engine confirms the claim
• False — the engine contradicts the claim
• Indecisive — the engine doesn't take a clear position

Why it doesn't count toward your score: The claims are stated in the prompt, so the conversation isn't measuring discovery. It's measuring accuracy — a different axis from visibility and recommendation.

What it's actually for: Fact Checker is most useful for brands where factual accuracy directly affects buyer trust or regulatory standing — security, financial services, healthcare, regulated industries. It also pairs with the Knowledge Base and Grounded Perceptions to form a complete accuracy stack: KB defines truth, Fact Checker actively tests whether answer engines confirm it, Grounded Perceptions show whether engines spontaneously match it.

Content implication: Update product/pricing/certification pages so the right facts are public and crawlable · Correct third-party listings · Publish authoritative reference content for high-risk claims

See Fact Checker Agent for the full feature page.

Quick Reference Summary

Two agents measure. Three agents analyze. Prompter and Recommender generate your KPIs. Introspector, Comparer, and Fact Checker help you understand the narratives, competitive dynamics, and factual accuracy behind those numbers. All five are essential — but they play different roles.

Next: From Data to Content Strategy