Einstein bots are part of Salesforce Service Cloud. The main purpose of Einstein bot is to interact with customers quickly and accurately without waiting for a human agent. Einstein bots can send messages, ask questions and perform the actions based on the rules defined or based on the customer input.
Let’s take an example where we have configured a bot to help the end-users to track and manage their orders. Let’s say I ordered myself a brand new BMW and have asked for home delivery. Now I’m being impatient and want to know when my car will be delivered.
I ask the bot “Tell me the status of order O-00123456” or “Where is my O-00123456”. The bot replies with the status along with some other information. In order to achieve this, we need to create Einstein Intents Sets. So let’s see what Intent means: – Intents are basically training datasets that contain the many ways in which your customer may ask the bot about one particular thing. This means we tend to configure the Intent of a certain request.
To create this go to Setup, use the Quick Find box to find Einstein Intent Sets and create a new intent set called Order Status Intents. After you’ve created the intent, edit this intent set.
From the Intents section, create at least two intents: Main Menu and Order Status.
Once you click on edit you would see an option to add Utterances. The Utterances are the unique possible customer inputs like:
- Can you check the order status for me?
- Did I place the order successfully?
- Order hasn’t arrive yet
- When will it be shipped?
Please note you would need to provide a minimum of 20 (provide as many as you can) Utterances. Most importantly a certain Intent should contain Utterances only for the intended purpose.
Now once you have configured the Intents and Utterances. It’s time to associate these to the bot dialogues. For example, we have created a dialogue named Order Status. This dialogue should wait for the customer’s input (not recommended but for our example).
In the above example, we have associated the Order Status Intent with the relevant dialogue. This is done so that when the user provides the input the bot call the dialogue to perform specific actions.
Please note: If you have not yet enabled dialog intents for this bot, you’ll have to click the Enable Dialog Intent button beside the dialog name before you can access the Dialog Intent subtab.
Next is to train the bot, now go to the Model Management tab and click the Build Model button. This will help the bot to understand the user inputs and redirect to the intended dialogue.
After all the above steps, finally we want the bot to respond. When the bot analyses what is the intent and which dialogue it is referring to, we can add our own logic in the dialogue to make it super responsive.
For example, if the bot calls the Order Status dialogue after customer input, we can ask a question using the dialogue (for example to provide the order number). Using this information we can further configure the dialogue action such as call apex, flow, object search etc.
Finally, this is how it will look like