By the end of this guide you’ll have a working AI Workflow that can greet visitors, grab their details, and—if they ask—book them straight into your calendar. We learn best from examples, so the best way to start learning an overview about workflows is by building a simple example.
Create a Workflow
Navigate to the Agents tab and click the + to create a new AI Workflow for your Agents. Name the workflow, optionally add the workflow to a workflow group, and pick the type of workflow this is (must pick a source type that cannot be changed later).
Connect a Source
At this point you may pick a source that you would like to connect to this Workflow. You can do this later if you wish as well. After connecting a source, you can choose which channel and tag rules you want your AI Agent to follow to be ON or OFF on contacts for each source.
Pick How to Get Started
Next, you can choose how to start creating your AI Workflow. You can create a workflow from scratch, from a template library, from an AI prompt or from the marketplace (coming soon).
Option | When to use it |
Template | Quick start with a pre‑built flow. |
AI Prompt | Tell CloseBot what you need in plain English; it builds the flow for you. |
Scratch | Drag‑and‑drop everything yourself (what we’ll do below). |
Marketplace is coming soon—share & sell your own flows.
Drag‑and‑Drop the Core Actions
In this example, we use some of the core actions of a workflow to build a rather robust Agent Workflow.
This workflow will allow your agent to conversationally get a contact's first name and email before just hanging out to answer their questions and engage. However, if at any time the contact wants to book a demo, the Agent is free to schedule them for a demo call before continuing once again to converse.
Nodes used in this example:
Objective - gather information from a contact and optionally update a field
Conversation - hang out and chat
Custom Scenario - jump to another part if something is detected
GHL Booking - conversationally book to a HighLevel calendar
Tip: The Agent will always start at START and continue to follow the arrows between nodes
Objectives
We want the Agent to immediately collect the contact's name and email address. We can do this within a single objective node so the Agent is more likely to ask for this information as one message instead of first asking for one, getting an answer, then asking for the other.
Conversation
Most of the time, after collecting information, businesses want their agent to be there to answer questions and chat back and forth. Whenever you want the Agent to hang out and chat like this, use a conversation node like we've done after our objective node.
Custom Scenario
It's common for businesses to want their Agent to jump to do something specific if it detects that it should do so. A common example is when we want the Agent to answer questions and chat freely, but jump to conversational booking if it detects the contact wants to book a demo.
GHL Booking
Once the Agent detects this scenario, it jumpt down from the START main flow and into the custom scenario flow. Now we have to decide what we want to happen when it detects that someone wants to book a demo. Naturally, we want it to book the demo with the GHL Booking node.
Note: You will be able to select the calendar from a dropdown of options or input your Calendar ID from the HighLevel calendar.
Something Cool About Workflows
When the agent is finished with the custom scenario sub-flow, it will automatically return to the point in the conversation where it was previously on the main START flow. However, you can also tell the flow where to jump back in as shown here.
Attach a Persona & Publish
Now it's time to attach your persona. Without a persona attached, you won't be able to Publish and test your Agent Workflow.
Other Settings
You'll also see additional settings for Bot Information where you can give the Agent context about the business it works for and why the conversation is happening.
Next Steps
Once you have your Agent Workflow and connected Persona finished, you can click to Publish and Test your Agent! Continue to tweak your agent until you're happy, upload items to the knowledge library for your Agent to use when chatting with contacts from your source, and you're ready to go!