Skip to main content
As you may have seen in the previous chapter about workflows, cogs are the individual instructions that make up an Engine’s workflow. There is no secret: each cog represents a specific action, and they are executed in the order they are arranged in the workflow. There are a couple of different types of cogs, and we will keep releasing new ones in the future. In this chapter, we will go throuhg different types of cogs available in Cogfy Messenger and how to configure them.

​
Command cog

The Command cog is a special cog that allows you to trigger a differnet set of cogs when the user types the specified command name, preceded by a slash (/). The command cog. Commands are a great way to provide quick access to specific functionalities of your bot, without having to rely on LLM interpretation for intention recognition, which might not always be accurate. When a command cog is matched, the default workflow is ignored. For that reason, command cogs are placed outside the default section and they cannot be placed inside other cogs, such as routers.
Commands are shown to users as suggestions when they type the slash (/) character in the chat input.

​
Send Message cog

The Send Message cog is used to send a predefined message to the user. Because messages can be reused in multiple places and some of them can be quite complex (for example, Whatsapp Flows), we decided to move the creation of messages to a separate section called “Messages”, as you can find in the sidebar. After having at least one message created, you configure a Send Message cog to reference one of those messages in the cog’s dropdown, as shown below. The send message cog.

​
Run Assistant cog

The Run Assistant cog is used to run a Engine Assistant to generate a reply, considering the current conversation context. The run assistant cog.
Not to be confused with with the old Cogfy Assistants: Engine Assistants are a new concept of assistants that are scoped by Engine and meant to be used by Engines only.
Because of the complexity and reusability of configuring an Engine Assistant, we also moved their creation to a separate section called “Assistants”, which is covered in the Assistants chapter of this guide.

​
Router cog

The Router cog is used to route the workflow execution to different paths based on certain conditions. The router cog. Conditions can be configured by clicking on the pencil icon on the right of each path cog, which will prompt a modal where you can intuitively set up conditions, as follows: The router cog conditions modal. As long as there aren’t any typos, this path will only be matched if the first name is equal to King and the second name is equal to Witcher.
  • After executing the matching path, the workflow execution will continue after the router cog.
  • If more than one path matches the conditions, only the first matching path will be executed.
Cogs can be dragged and dropped from and into paths for better user experience.

​
LLM Router cog

Rather then relying on predefined conditions, the LLM Router cog is used to route the workflow execution to different paths based on the output of a Large Language Model (LLM), such as GPT, consdiering the current conversation context and any custom prompt you may want to provide. The LLM router cog.
Just like the Router cog, cogs can be dragged and dropped from and into paths.

​
LLM Router

The LLM Router cog on top allows you to define the model, some optional instructions to guide the model’s behavior and a default behavior in case none of the paths are matched by the LLM, as shown below: Edit LLM modal. The default behavior field describes what to do if none of the paths are matched. You can choose between the following options:
  • Use first route: Route to the first path defined in the LLM Router cog.
  • Use last route: Route to the last path defined in the LLM Router cog.
  • Throw error: Throw an error and stop the workflow execution.
  • Stop: Stop the workflow execution. Further cogs will not be executed.
  • Run next cog: Continue the workflow execution with the next cog after the LLM Router cog.
  • Send default message: Send a predefined message of your choice to the user.

​
LLM Router Path

Each LLM Router Path cog allows you to define the name of the path and a prompt that will be used to determine which path to take. Both the name and the prompt will be used by the LLM. Edit LLM Router path modal.
Keep the route names short and simple, as they will be provided to the LLM as options to choose from. Excessive complexty may lead to LLM confusion and unexpected behavior.

​
Update Contact cog

The Update Contact cog updates a field of your choice in the contact profile. The update contact cog.
For now, the only supported options are adding and removing tags from the contact, because it wouldn’t make sense to update other fields to predefined values. In the future, we may add support for more complex updates using variables.

​
HTTP Request cog

The HTTP Request cog allows you to make HTTP requests to external services. The HTTP request cog. When you click on the pencil icon on the right, a modal will open where you can configure the HTTP request in detail. Most fields are self-explanatory, and you can also configure a request timeout for when the external service takes too long to respond, as you can see in the example above.