---
title: "Chatbot or agent: which steps remain | StudioChat"
description: "The same message put to a button flow, an assistant and an agent, with what each one does and the steps that are left over for a person to finish."
url: https://studiochat.io/which-steps-remain
locale: en
translations: { es: https://studiochat.io/que-pasos-quedan.md, pt: https://studiochat.io/que-passos-ficam.md }
site: https://studiochat.io
---

Tool

# The same message, put to all three

One customer message, put to the three things sold under the same name: a button flow, an assistant that answers and an agent that executes. Each one shows what it does, one action at a time, and the steps a person is left with afterwards.

The same message, put to all three

Pick the message that comes in every day:

I want to change the delivery addressSend me September's invoiceI need 20 boxes of the 480 for ThursdayRunning late today, I don't leave until 10

Button flow

Shows a menu of options and none of them is changing the address.

what it does, step by step

* Shows the menu it has loaded
* Cannot find the option and offers a person
* left for the customer: 3
* left for the team: 2

5 steps pending

Assistant that answers

Correctly explains how an address gets changed and sends the link to the form.

* Searches the knowledge base
* Writes the steps out and sends the link
* left for the customer: 2
* left for the team: 1

3 steps pending

Agent that executes

Changes the address and sends the confirmation with the new destination.

* Identifies the account from the number writing
* Finds the order in the system
* Checks it has not shipped yet
* Writes the new address onto the order
* Notifies the carrier of the change
* Sends the confirmation to the chat
* left for the customer: 0
* left for the team: 0

0 steps pending

With the same message there are 5 steps pending in the first case, 3 in the second and 0 in the third. What changes is not the wording of the reply: it is whether anything touched a system.

The three types are categories of behaviour, not products. Any platform can land in either column depending on the case, which is why the test gets run on your messages and not on the demo's.

This object is the interactive half of an article: [Chatbot or agent: the same message put to all three](/blog/chatbot-o-agente-que-estas-comprando)

## The playbook already exists. We multiply it.

[Talk to us now](https://cal.com/studiochat)

[Book a demo](https://cal.com/studiochat) [Or write to hey@studiochat.io](mailto:hey@studiochat.io)