❓ Frequently Asked Question or Initial Context
Sometimes, it is observed that a conversation assigned to an agent returns to bot control without direct intervention. This behavior often causes confusion when the agent is expected to maintain full case follow-up.
🛠️ Troubleshooting: Step-by-step to identify the problem
Review bot configuration
Verify if the bot has rules that re-evaluate active conversations.
Confirm if there is conditional logic (keywords or inactivity times) that causes it to take control again.
Check assignment logs
In the conversation history, review which entity performed the reassignment (was it the bot, the system by a global rule, or a manual move?).
Verify automation flows
Make sure there is no active automation that re-enters the customer into the bot flow if it does not detect a quick response from the agent.
Agent status
Confirm that the agent is online and in assignable status. If the agent goes to "Away" or disconnects, Atom may return the conversation to the bot according to your contingency settings.
Review group membership
The agent must necessarily belong to the exact group to which the bot transfers conversations.
✅ Possible solutions
Modify the bot flow: Ensure that after the transfer, the flow ends correctly and there are no "open nodes" that could be reactivated.
Adjust inactivity rules: Review if you have configured automatic reassignment to the bot when the agent does not respond within a certain time. E.g., Automated tasks.
Add a conditional wait: Introduce a "wait" block after the assignment to give the agent room to maneuver before any bot re-intervention.
Review team configurations: Clearly define assignment hierarchies to prevent the bot from "interfering" with already-initiated human work.
📝 Additional Notes
Some behaviors may be linked to system updates that require specific technical review.
Reassignments can be influenced by business hours settings: if an agent receives a chat outside of hours, the bot could take back control to send an away message.
💼 Practical Example / Use Case
Scenario: The "Attention 2" bot transferred chats to agent Pérez after detecting the word "training". However, the chat returned to the bot almost immediately.
Diagnosis: The bot flow had no "End" node after the transfer. Not detecting an instant response from the agent, the bot interpreted that it should continue assisting the user.
Solution: The flow was adjusted by adding a logic closure and a 5-minute wait before the bot could attempt any new automatic intervention.
📞 When to contact support?
Contact support if:
There is no visible rule that justifies the reassignment.
The behavior is intermittent and does not follow clear logic.
The problem persists after correctly configuring flows and statuses.
Please include:
Screenshots of the bot flow.
ID or history of the affected conversation.
Name of the bot and agent involved.
Don't let the bot interrupt your team! 🚀
If you notice that a chat returns to automatic control after being assigned, review that your flows have a clear "End" node and that there are no overly aggressive inactivity tasks. Make sure your agents maintain their "Assignable" status and adjust contingency rules so that the human has full control once the bot has accomplished its mission. ✅