In Genezio, a conversation is one run of a scenario with one answer engine. In a conversation, Genezio sends prompts to an answer engine and records the results for the analysis. Thus, each conversation is an instance of a scenario that Genezio runs with one answer engine.

​
Scenario and conversation

The relation between a scenario and a conversation is simple:
  • A scenario gives the situation and the goal of the persona.
  • A conversation occurs when Genezio runs that scenario with an answer engine.
For example:
  • Scenario: “A startup founder looking for a CRM to manage early leads”
  • Answer engine: ChatGPT
When Genezio runs this scenario with ChatGPT, it makes one conversation. When Genezio runs the same scenario with more than one answer engine, it makes more than one conversation. For example: Each of these conversations can give different responses, citations, and brand mentions.

​
Single-turn and multi-step conversations

The structure of a conversation depends on the type of the Genezio agent.

​
Conversations of the Prompter Agent

For a topic of the Prompter Agent, the conversation has one message. Genezio sends the prompt that you wrote word for word to the answer engine. Example prompt:
User query: What CRM should a startup use?
Genezio records the response of the answer engine and analyzes it.

​
Multi-step conversations

For a topic of the Recommender Agent, the Introspector Agent, or the Comparer Agent, the conversation is multi-step:
  1. The scenario gives the initial situation.
  2. Genezio makes the prompts during the conversation.
  3. The conversation can have many turns in the two directions.
These conversations simulate how a real user explores a topic with an AI assistant.

​
Conversations and answer engines

Genezio runs conversations with many answer engines. Examples:
  • ChatGPT
  • Claude
  • Gemini
  • Perplexity
  • Google AI Overview
When Genezio runs scenarios with different answer engines, it can measure how the visibility of a brand changes from one platform to another. To select the answer engines, refer to Selecting answer engines.

​
Special case: Google AI Overview

Google AI Overview operates differently from a conversational AI assistant. AI Overview does not start from a conversational prompt. It starts from a text that is almost the same as a web search query. Because of this difference:
  • Genezio changes the prompts into search-style queries for AI Overview.
  • These queries are almost the same as the text that a user types in the Google search box.
Example of a conversation prompt:
User query: What CRM should a startup use?
The AI Overview query for the same prompt:
AI Overview query: best CRM for startups
Genezio makes these queries automatically. But users can override the query that Genezio makes. Do this to test a specific search text.

​
Conversation data

Each conversation gives a structured set of data that Genezio analyzes. This data includes:
  • The prompts that Genezio sent to the answer engine
  • The responses of the model
  • The brands that the response mentions
  • The cited sources
  • The follow-up searches (query fanouts) of the answer engine
This data is the base for metrics such as AI Recommendations, AI Visibility, share of voice, and competitive analysis.

​
Conversation detail view

Genezio has a conversation detail view. In this view, users examine the results of one conversation. This view shows:
  • The full conversation between Genezio and the answer engine
  • The prompts of each turn
  • The responses of the model
  • The query fanouts that Genezio extracted
  • The citations that Genezio found
  • The brands and competitors that the response mentions
  • The products that the response mentions, tagged and highlighted in the text
With this view, teams can see exactly why a brand appeared, or did not appear, in a response of an answer engine. The transcripts know the products. Genezio highlights the products that an answer names, and it shows tables and maps in the text. Click a product card in the transcript to open the Product Details drawer of that product. You stay in the report, and you do not go to the page of the answer engine. The card does nothing if your brand has no shopping report or if the conversation did not record the product. Thus, you can find the moment when an answer engine recommended a specific product or ignored it. Refer to Product visibility.

​
Search the conversations and check a claim

The conversation search page has two tabs. Use the Filters of the report to limit both tabs to the topics, personas, and answer engines that you care about. Search finds what the answers say about a subject, such as “our pricing” or “credit for young people”. It also works for short searches of one or two words. Genezio matches the meaning of your words and the exact words together, and it puts the best passages first. Each passage shows how well it matches: strong, possible, or weak. Check a claim tells you how many answers state something about your brand. Type the claim, for example “Our plans start at $49”. Genezio reads the conversations and shows:
  • the passages that support the claim,
  • the passages that contradict it,
  • the number of conversations and passages for each group,
  • the conversations of each answer engine, with its logo.
Why this is important: you can measure if a wrong claim spreads or if a message that you pushed reaches the answers. Use it before and after a content change to see the effect. A claim check can take some time on a large report. You can Stop it and Continue it later, and you can start a search in the other tab while it runs.

​
Why conversations are important

A conversation is the core unit of analysis in Genezio. Genezio calculates each visibility metric, each insight, and each competitor analysis from the results of many conversations. These conversations cover many topics, personas, and answer engines. Genezio analyzes these conversations at scale. Thus, it gives a measurable view of how answer engines show brands.

​
Next steps

To learn how Genezio structures and analyzes the information from conversations, refer to:

​
Conversations in the public API

The public API reads conversations. Refer to List conversations.