You can create a Freddy AI Agent or build a chatbot without requiring custom development with the chatbot builder. We’ll walk you through the process of configuring your first chatbot in the following steps:
- Create a Freddy AI Agent
- Create a chatbot
- Create Flows
- Set up conditions
- Trigger Actions
- Preview the chatbot
- Deploy the chatbot
- Unpublish the chatbot
- Analyze the chatbot behavior
- Sorting options to find your chatbot
Create a Freddy AI Agent
Freddy AI Agents are people-first AI-powered bots that learn from various knowledge sources to answer how-to queries. The AI Agents can learn from Files, Web links, Solution articles, and custom Q&As.
Freddy AI Agents can:
- Offer customized responses to customer queries from your knowledge sources in just a few clicks.
- Provide an enhanced conversational user experience with multi-turn question answering, enhanced small talk, a multilingual experience, and much more.
Learn how to set up your first Freddy AI Agent and get started with automating your customer service effortlessly.
Create a chatbot
Follow the steps below to create your first chatbot.
- Log in to your account and click on the AI Agents and Chatbots icon.
- Click Create new.
- Enter the Name, select Chatbots option, and choose the Primary language and your preferred Channel.
- In the Knowledge Sources for your bot pop-up that appears, choose Learn from FAQs > Create New. This is an optional step as you can toggle off Learn from FAQs and proceed to create the chatbot. You can enable this later if required by going to Natural Language > Configure > Natural language settings > Enable Learn from your FAQs
- You will be redirected to the Flows tab, where you can start building flows for your chatbot.
- Alternatively, you can create chatbots by cloning an existing published chatbot. To clone a chatbot, click on the three dots next to the chatbot name, and click Clone.
Create Flows
A flow is a series of connected dialogs that make up a chatbot. You can have any number of flows in the chatbot that get executed sequentially when the chatbot is triggered.
When you create a new chatbot, you will be automatically redirected to the Flows tab, where you can see sample flows created to help you get started.
To create new flows:
- You can also create a new flow by cloning an existing flow. To clone a flow, hover over the flow and select Clone.
- Start adding dialogs to the flow. You can add any number of dialogs as required to a single flow and include either a message or an action. For more details, check Configuring dialogs in the bot builder.
Note: You can delete a dialog if it's no longer required. However, you can only archive a flow and not delete it.
Set up conditions
- All the dialogs created are executed subsequently unless a condition in one dialog redirects the customer to a different dialog or if there is a break in the flow. If you’re using the button input type or carousel input type, you can trigger actions corresponding to received responses.
- Similarly, you can redirect customers to a particular dialog in a different flow based on their choices. You can also use the nested AND/OR operators for these evaluations.
For example, a travel agency can choose to display destinations to plan trips for their customers. In another example, a bank evaluates the customer’s age and then accordingly shows interest rates.
Trigger Actions
You can also execute actions with the chatbot. The Actions feature in the flow allows you to execute events based on your requirements.
For example, you can trigger an API as part of a dialog to create a ticket in Freshdesk and store the ticket ID as part of the API. In that case, you can use the Trigger API action to fetch the ticket ID from the API parameters and show it to the customer while they are interacting with the bot.
To add an action in a dialog, click on the plus icon within a dialog and choose New Action, and then choose the action type
The different action types you can add are:
- Dynamic actions: Trigger API, Set property, Transfer to Answers
- Agent handover: Transfer to Agent
- Conversation actions: Start a new conversation, Stop the conversation, Resolve the conversation
- Widget actions: Trigger actions on widget, like triggering an event when the widget opens.
For more details on triggering actions, check Configuring Actions in the bot builder.
Preview the chatbot
Once you complete the chatbot configuration, click on the Preview button on the top right to preview the chatbot. The preview option lets you visualize the connections between the different flows with the Holistic view button on the bottom left.
Note: When previewing a chatbot, you can use the show logs button to switch to the activity logs to understand your chatbot performance, such as the conversation ID, API calls, and where the flow breaks, for example.
Deploy the chatbot
Once you've finished setting up the bot, you can publish and deploy your bot. When you deploy your bot, you will be automatically redirected to the Topics page, where you can choose the topic on which you want to deploy the bot. Check out this article to learn more about the mapping your chatbots to various supported channels.
Unpublish the chatbot
If you feel like temporarily deactivating the chatbot, you can use the unpublish button, which will instantly deactivate the chatbot with a click of a button.
Note: Unpublishing a chatbot version does not allow you to edit the flow, you would have to create a new version and ensure the chatbot is in draft version to edit the flow.
Analyze the chatbot behavior
In some situations, a chatbot may face an error, such as when an API fails, or a function is incorrectly executed because of an improper input. These errors usually arise at a dialog level. Admins must quickly identify where the error appears to identify, troubleshoot, and fix the issue to offer a customer service experience free of technical issues and glitches. For more details check, Troubleshoot bot errors with the new dashboard.
Sorting options to find your chatbot
Admins can sort chatbots based on creation date, bot name, or version count, making it simpler to locate and manage specific chatbots.