This guide collects real-world problems our customers have run into with AI Agents, along with the setup that solves each one. Each recipe works on its own, so pick only the ones you need. They're ordered from simplest to most advanced, and at the end we show how to combine them for more complex logic.
1. Manually Hand a Contact to the AI Agent
The problem
A teammate wants to hand a specific contact to the AI Agent on demand, without waiting for automatic enrollment rules to match.
The solution: an enrollment tag
In Contact Enrollment, add the condition Tags → contains → To Support.
To enroll a contact, add the To Support tag from the contact sidebar, the Contacts page, or a Workflow.
Result: Any teammate can route a contact to the agent with a single tag.
Tip: Add a Remove tag action for To Support in Success and Fallback Actions so the tag can be reused the next time you want to enroll the same contact.
2. Don't Enroll Contacts Your Team Is Already Talking To
The problem
A contact is already in an active conversation with someone on your team. When they reply, the AI Agent enrolls them and interrupts the conversation.
The solution: check recent outbound activity
In Contact Enrollment, add the condition Last Message Sent → not updated in the last 1 day.
Result: The agent only enrolls contacts that nobody on your team has messaged in the last day. If a teammate is actively working with the contact, the agent stays out of the way.
3. Prevent Re-enrollment Right After the Agent Finishes
The problem
The AI Agent completes its session and runs all Success or Fallback actions. The contact then replies with something like "Thank you", which matches the enrollment criteria again, and the agent re-enrolls them and starts a brand-new session.
The solution: a cooldown tag
Give the contact a temporary tag when the agent finishes, exclude that tag from enrollment, and remove it automatically after 1 day.
Tag the contact when the agent finishes. In your AI Agent, add an Add tag action (e.g. AI Engaged) to both Success Actions and Fallback Actions.
Exclude the tag from enrollment. In Contact Enrollment, add the condition Tags → does not contain → AI Engaged.
Remove the tag after 1 day. Create a Workflow triggered when the AI Engaged tag is added, add a 1 day delay, then a Remove tag step for AI Engaged.
Result: Follow-up replies like "Thanks!" are ignored for 24 hours. After that, the contact becomes eligible for enrollment again.
Tip: Adjust the delay to match your use case, e.g. a few hours for high-volume support or several days for sales outreach.
Combining Recipes for More Complex Logic
These recipes work together. For example, a support agent using all three could have two enrollment groups:
Group 1 – Automatic enrollment (all conditions must match)
Your usual trigger (e.g. inbound message received)
Last Message Sent not updated in the last 1 day (Recipe 2)
Tags does not contain AI Engaged (Recipe 3)
OR
Group 2 – Manual enrollment
Tags contains To Support (Recipe 1)
Plus a Workflow that removes AI Engaged 1 day after it's added.
Result: The agent picks up new inbound conversations on its own, steps aside when your team is already engaged, doesn't restart after a "Thank you", and can still be triggered manually at any time.
Need Help? ✋🏻
Contact us on live chat or send an email to us at [email protected].