Monthly Archives: May 2019

Automate Tier 1 while Keeping Agents in the Loop

Using Live Chat to Seamlessly Handoff Chats to Agents.

Chatbots are useful at *Tier 1*, but humans are good at solving complex problems. Sometimes it is tiring to do the same thing (e.g., booking a hotel room) over and over again while keeping the quality of service high.

Now we have made it easy to argument chatbot experience with Live Chat.

I will head to the Smartloop Conversation Builder and create a Nespresso Cafe bot.

Let’s assume, it is a full-service bot to help me order and track Nespresso pods. I can track my shipment or in case If I never received it, it can seamlessly transfer me to a live agent.

The basic part is simple, I can just say “chat with an agent “ and it immediately takes me to the agent flow or if the bot does not understand me, it can try possible routes and eventually take me to an agent too. This is more or less a happy path. What if the shipment is lost, Tier 1 should not be used to solve this type of problem but an agent should be there to solve it for me and resend or refund the order if necessary.

Here in this scenario, I’ve bought a VirtuoLine pod, and it should not send me to the original line support channel. The bot should be collecting necessary profile information based on my previous engagement and should send me to the correct support channel (this is where AI part kicks in). Here in this case, when I type “I’ve lost my order”, it immediately takes me to the correct VirtuoLine agent which is set in the “new-order” block as a user attribute while the bot is taking me to the alleyways of conversation

Now as the agent is connected, the bot is paused and communication happens between the agent the messenger seamlessly.

The agent should be able to check the history and user should be able to return to the bot either way by typing “exit” as configured in the Live Chat plugin or agent can end the conversation.

Here, AI and Human are complementing each other, along with flows defined as per our need to streamline the support process. Moreover, as the number of subscribers grows, we should be able to see their activity and jump in / out a conversation as soon as we see the conversation goes off track. This will help answer “Why most bots are failing”. We should not just build it and go to the bar while frustrated users *ditch* the service. It is a constant learning process and machine learning is useful and can do more for us only then.

Live Chat and ability to seamlessly handoff to an agent is available now for Facebook, Viber, Telegram and Web.

Get started with Smartloop Platform for Free and send us your query at


Automate Tier 1 while Keeping Agents in the Loop was originally published in Smartloop on Medium, where people are continuing the conversation by highlighting and responding to this story.