Guide

What is a conversational agent (and how it differs from a flow bot)

A conversational agent is an AI program that talks with a person in natural language, understands what they want, checks the company's information and acts to resolve it: it answers, qualifies, schedules, opens a ticket or brings in someone from the team. It does not follow a fixed script of buttons. It interprets the message and decides the next step within rules set by the company.

Data checked on October 12, 2026

The difference from a flow bot

For years, automated support was built with flows: a menu of options, where each click leads to a branch drawn in advance. It works as long as the customer asks what the designer anticipated. When they write something different, the flow gets stuck, repeats the menu or hands off to a human.

The conversational agent changes the logic. Instead of drawing every path, the company describes the goal, the tone and the limits, and gives the agent the content it needs. The person writes in their own words, and the agent understands.

How a conversational agent works

• Understands. Reads the message, the context of the conversation and, when available, images sent by the person, such as a screenshot of an error or a photo of a product.
• Checks. Finds the answer in the company's knowledge base: documents, website pages, spreadsheets, catalog, support rules.
• Acts. Does what the goal calls for: qualify a lead, book a time slot, open a ticket, look up an order.
• Hands off. When a case needs a human decision, it transfers to a person or to another agent, carrying the conversation history.
• Measures. Records every step, from the first message to the conversion, so the company knows what worked.

Where it is used

The same mechanism serves different goals, and each one calls for its own setup:

• Support and customer service: resolves the case, applies the company's rules and generates a ticket.
• Lead capture: talks with the visitor and delivers a qualified lead, with the reason for their interest.
• Sales and purchase intent: recommends products from the catalog and looks up orders.
• Scheduling: books the time slot straight into the calendar, with no back and forth.
• Reception and multi-agent teams: understands the request and routes it to the right agent.
• Self-service and FAQ: answers repeated questions based on the company's content.

What to watch for when choosing a solution

A well-built AI agent solves a lot, but a poorly configured one does damage. These are the questions worth asking any vendor:

• Where do the answers come from? The agent should answer based on the company's content and admit when it does not know, instead of making things up.
• Is there a handoff to a person? Every case that needs a human decision has to reach someone, with the history.
• Can you measure it? Without a record of the steps, there is no way to know how many conversations became customers.
• Which AI model is used and where is the data stored? The company has the right to know, and LGPD and GDPR require transparency about data processing.
• How much does it cost per use? Compare what is charged per conversation, per contact and per message, and what happens when the limit runs out.

Where CitraChat fits in

CitraChat is a platform for building and running conversational agents on your website and on the official WhatsApp API. Each agent has a name, tone and goal, is trained with the company's content and can work as a team with other agents. Every step of the conversation becomes an event for Google Ads, Meta, GA4 and other platforms. Plans start at $29 a month, with a 14-day free trial and no card required.

If you are evaluating alternatives, see the comparisons with Leadster, BotConversa and ChatSuite.

Frequently asked questions