Overview
Agent Node is a node that has its own instructions, using tools you've given it access to, until it is permitted to exit. There are 3 things to learn about the Agent Node.
Instructions → how to instruct your agent on how to act, what to do
Allowed Tools → how to give your agent access to tools it needs, when it needs them
Exits → when you want to exit a specific agent node, moving on to the next node that may have different tools and instructions
Instructions
Agent Nodes follow your global instructions set up within your Job Flow Settings, but you can also add specific instructions within each agent node. You can clarify how it should use tools, exits and fields.
Sections
You insert instructions by adding a new Section. Sections are organized texts for you to explain how the agent should behave. Sections can be organized however you want. Instead of typing a giant block of text, think about how you can break larger instructions into separate categories. Here's an example 👇
Agent Nodes, like real people, have an easier time reading text that organized neatly.
@ Mentioning Things
You can type the "@" character to pull up items for reference. Use this to mention variables, tools and exits.
@ → variables
@@ → tools
@@@ → exits
The Agent will do its best to use the tools and exits it has been granted access to, but explicityly mentioning how and when you want it to use these within the instructions can be rather helpful and improve reliability.
You will see an error if you try to @-mention a tool or exit that has been removed from this agent node
Tools
Tools are things your Agent Node can use to take action externally. Booking appointments, gathering availability, using custom tools... Without tools enabled your Agent Node is simply a chatbot that can do nothing. Learn more about tools HERE.
Exits
Exits are situations where you want to be able to exit from the Agent Node. A good example is a specific Agent Node you've created without access to the Book Appointment Tool when a lead is qualified to move onto another Agent Node that does have access to the Book Appointment Tool.
Exit Descriptions
Exit descriptions are one of two optional settings you can save to a created exit. This is where you can describe when the Agent decide to exit.
For example → use this when the lead has given all information we need
You don't have to use a description. You could achieve a similar effect by @@@-mentioning the exit within the instruction of the Agent Node Section(s).
Exit Tags
Exit Tags (or lists depending on source type), if set, will cause the lead chatting to immediately exit from this Agent Node if we detect the tag combination specified exists. With this, you can take action within your CRM automations that add or remove tags, exiting leads from their current Agent Node.
Example → Remove them from the QUALIFY agent node if they contain the tag "already booked"... maybe they booked an appointment already from your website.
If you have added exit tag filters and no description, and your Exit is not @@@-mentioned within the instruction sections, the Agent will never select the exit unless the tag rules become true.
Summary
The agent node has become the most powerful tool in your toolbox. It can be used to replace every other node that previously caused the bot to reply...
Objectives, statements, booking node, conversation node, etc. all can be made out of this one Agent Node. Learn more about tools for the agent node HERE.