Skip to main content

How do Flows by Classification work?

Learn how to automate workflows based on classifications in Atom. Connect the closure of your conversations with automatic actions to send surveys, reminders, or promotions without manual effort.

In Atom, you can configure automatic flows that trigger at the exact moment an agent classifies and closes a conversation. This functionality allows you to connect the end of human support with automation, significantly improving customer experience and reducing operational burden.


🎯 What is the objective?

Allow closing classifications to act as triggers within Flowbuilder. This way, a flow is automatically generated (such as sending a template or survey) when an agent classifies the conversation in a specific way.


🧩 Use cases by industry

  • Retail: A customer makes a purchase and the agent classifies the chat as "Sale closed". A flow with a satisfaction survey and discount coupon automatically triggers.

  • Insurance: The customer makes a "Technical inquiry". Upon completion and classification, a flow starts with informational resources and the option to schedule a call.

  • Education: An interested student is classified as "New inquiry". An automatic flow is activated that sends enrollment dates and financing options.

  • Automotive: If a customer files a complaint, the agent classifies it as "Complaint". This activates a retention flow that offers a free inspection or escalates to a specialized team.


👥 Who uses this functionality?

Role

Action on the platform

Administrator

Configures and designs flows in the Campaigns / Flowbuilder module.

Agent

Applies the classification when closing the chat in the Conversations module.


✅ Key benefits

  • Contextual automation: Precise flows are triggered based on the customer's actual reason for contact.

  • Greater efficiency: The agent quickly finishes the conversation and the system handles the follow-up.

  • Interaction without delays: The customer receives the promised information (or survey) immediately.

  • Reduction of operational errors: Human oversight is avoided by not relying on the agent to manually send closing information.


⚙️ How to configure it? (Step-by-step for Administrators)

  1. Go to the Campaigns module in your left sidebar menu.

  2. Select the Flows submodule.

  3. Click to create a new flow to enter the Outbound Flowbuilder canvas.

  4. Click on the initial node and select "Classification" as the trigger option.

  5. Select the specific classification or category that will activate this flow.

  6. Configure the waiting time. That is, how long the system will wait.

    to send the first message after the agent performs the classification.

  7. (Optional): Enable the "Option to disable flow sending". By default the flow will always be sent, but enabling this gives the agent freedom to skip the automation if the context of that conversation doesn't warrant it.

  8. Continue adding components to your flow (message template, conditionals, etc.).

  9. Click Publish.


💬 What happens on the Agent's side?

  • The agent converses normally with the customer from their inbox.

  • When finishing the interaction, they select the corresponding closing classification.

  • If a flow is linked to that category, the system shows them an automatic notification.

  • If the administrator enabled the permission (Step 7), the agent can decide whether they want the flow to be sent or if they prefer to cancel it for that user.

  • The system will record in the conversation history both the classification and the action taken (flow sent or flow omitted).

💡 Configuration tip: If you don't yet have your closing categories set up, we recommend configuring this first. Consult internally the guide on "How to create and edit classifications in Atom".


Take advantage of every conversation closure! 🚀

By connecting your classifications with Flowbuilder, you guarantee perfect follow-up, nurture your customers, and free your team from repetitive tasks. Make every chat end with a great automated experience. ✅

Did this answer your question?