First, let me explain a scenario to quickly understand how Einstein service replies for email in Salesforce Agentforce works. Then I will explain how we can achieve this in Salesforce.
So, let’s say you are a support engineer and the customer has a question about the Wi-Fi router not working. Now, to get the proper information related to the user’s query, you will go through the related knowledge article that you have.
Once you find the correct information, you type out the answer manually and send it to the customer. Isn’t it lengthy to go through all the information and respond to the customer’s query?
We can implement AI automation using Einstein in Agentforce to simplify the process. Now, further, I will explain how we can set up to automate this process using Einstein service replies.
Einstein Service Replies for Email in Salesforce Agentforce
The Einstein Service reply features also help agents quickly draft emails from the knowledge article. We know that the knowledge recommendation feature in Service Clouds recommends all the relevant knowledge articles right on the case detail page, and agents can use that to resolve a case.
To help agents, send an email to the customer Einstein for replies that can help generate the email quickly from the knowledge articles.
Once it is enabled, the agent can click on the drop-down and select Draft Einstein email, and Einstein will generate the email from the knowledge article, which agents can edit and send the email quickly to resolve the case.
Einstein automatically reads the customer’s message and understands the problem. It quickly checks the related knowledge articles for you and finds the most helpful information.
Then, it gives you short and ready-to-send reply suggestions, so you don’t have to search or type everything yourself.
Set up Einstein Service Replies For Email in Salesforce Agentforce
Now, we will understand how to enable Einstein service replies for email, which helps agents draft email content quickly from the knowledge articles displayed on the recommended articles on the case detail page.
I have created a new Salesforce developer org because Salesforce launched a new Developer Edition on March 3rd, 2025, which now comes complete with Agentforce and Data Cloud.
Now, we need to enable Einstein AI. So, let’s go to setup. Search for Einstein Setup and Turn on the Einstein option.
After that, we need to enable Agentforce. Again, search for the Agents and Enable the Agentforce, which is the default agent for Salesforce, by clicking on the toggle button.
Enable Knowledge User and Article in Salesforce
The Einstein service replies that they utilize existing knowledge articles to draft the email. For that, we need to enable and load a few knowledge articles in this org so that they can appear in the recommendations section.
Now, when searching for knowledge in the setup, it does not appear in the search results. This is because to work with knowledge, our user needs to be a knowledge user.
To do that, navigate to the users, select the user you want to become a knowledge user, and check the checkbox.
After enabling the knowledge user, we need to activate knowledge. To do that, search for the Knowledge Setting in Quick Find.
To enable knowledge, check the checkbox, click Enable Salesforce Lightning Knowledge, and click Okay to proceed. So now knowledge is enabled in this org.
Now, we need to edit some knowledge settings, as given below. To do that, click the Edit button and enable the following settings.
Now, check the Activate Validation Status field checkbox. Then, Enable Lightning Knowledge so that the knowledge article appears in the Lightning UI as well.
Next, make this article available to the communities, which are internal apps, customers, and partners.
Next, this is the most important point: Suggest related articles on cases. This checkbox ensures that the recommended knowledge articles start appearing on cases based on the field you define.
Right now, you see that the Subject field is selected by default, which means that when Salesforce recommends knowledge articles, it will look for the content of the subject.
Then, it will find that in all of the knowledge articles, whichever article it finds relevant will start showing up in the recommended knowledge articles section.
Then click the Save button to apply the changes.
Insert Knowledge Articles in Salesforce
As of now, we have enabled Agentforce and knowledge settings, and the next step is to load some knowledge articles into this org. So, for that, I have one Excel file with a few knowledge articles.
We only need the title, summary, and URL name to create the knowledge articles. So copy these, then go to the knowledge article again and click the Inspector button.
There, you will have the Data Import operation to insert the knowledge articles.
Now, paste the copied content in the Data option. Then, the Object where knowledge articles are stored. So here, we need to select Knowledge__kav.
In the Field Mapping, you can see that the external file columns and internal fields are mapped. These are the three mandatory fields for creating a knowledge article.
After that, click on the Run Insert button to create the knowledge articles.
Again, click the Insert button to see that the knowledge articles are inserted into this system.
But it needs to be published to work with knowledge, so if I open any record by clicking on ID, you will see it is in draft mode.
You can see that the Knowledge object tab is still in draft mode, and all the knowledge in draft mode does not show up in the recommended section.
To edit this object, click the Gear icon and the Edit Object option.
Then go to Page Layout -> Mobile & Lightning Actions -> We want to drag the publish button to the UI, so we have to enable this lightning action.
After that, drag and drop the publish button into the page layout.
You can also drag other buttons. For example, you have Delete Draft or Delete Article. Then click the Save button to apply the changes.
Again, as you navigate to the knowledge tab and refresh the page, you will see the buttons that you added to the page layout. Now, we have to publish draft articles by clicking on the Publish button.
When you click the publish button here, you need to select Publish now and then click the Publish button again.
The article’s Publication Status has changed from Draft to Published. Thus, this organization now has a few knowledge articles, one of which is published.
If Salesforce finds that this knowledge is relevant to the case we are working on, this article should appear in the recommended section.
Enable Einstein Service Replies For Email
Now, let’s move to the next step of enabling Einstein Service replies to email. For that, in Quick Find, search for Einstein Service. You can see Einstein service replies for email. Let’s open this, and we have to turn this on.
So let’s click on the toggle button, and now service replies are turned on.
But to generate the content from the knowledge article, we have to ground the data from the knowledge as well as the case object. So, for that, click on this button to Set Up Grounding.
So, turn on this Service AI Grounding. This is useful while generating the content for the email and chat features. We need to tell Einstein which field to use on the knowledge article and on the case object to ground the response.
For that, as you scroll down on the Knowledge, click the Edit button. Here, we are saying that the Title should be found in the title field of our knowledge article, and the Summary should be found in the summary field of the knowledge article.
Then click on the Save button. Also, you need to enable the Grounding with Knowledge option.
Similarly, click on Edit for the Case. Here, we can say that the Subject can be found in the subject field and the description in the description field.
Then, you can click on the Save button. You also need to enable the Grounding with Case option.
Now, let’s go to the Service app and see if these knowledge articles are coming up on the case. To do that, click the App Launcher and switch to the service app.
Open any case. On the left side, the suggested knowledge article section will appear. There, you will find articles related to your case Subject.
If you don’t have any knowledge articles related to the case subject, you will not find any articles there.
But when you click on the drop-down, you will not see the Einstein Draft email feature enabled yet because this user needs to have one permission set.
So, let’s go to the logged-in user and assign the permission set, which is Email response user.
So, go to this user, go to the permission set assignment, and assign this permission set.
Here is a search for the email response user. We need to assign this permission set to this user, so let’s click on Save.
After that, we need to enable the Email-to-Case feature to send the emails.
So now the agent can write the email using normal email writing and then send it, or they can use this knowledge article to generate that email for them.
So let’s just utilize this and click on Draft Einstein Email.
Here, you can see that the agent generated this email. If you see the content of this email, it mentions the name of the contact for this case. And then it generated this content.
This was generated based on the knowledge article. Again, if the agent wants to change this content, they can do that.
Otherwise, you can directly send this email.
Proof of Concept
Here you can see the customer received the email with the steps to resolve his query.
In this way, we can set up and use the Einstein service replies for email in Salesforce Agentforce to make the support agent’s task easy and accurate.
Conclusion
I hope you have got an idea about how Einstein Service Replies for Email in Salesforce Agentforce works, and then I have explained how we can achieve this in Salesforce.
- Create Agentforce-Enabled Scratch Orgs From Salesforce Developer Edition
- Prompts and Prompt Builders in Salesforce Agentforce
- Agentforce for Developers in Salesforce
- Introduction to Model Builder in Salesforce Agentforce
Shubham is a Certified Salesforce Developer with technical skills for Building applications using custom objects, approval processes, validation rule salesforce flows, and UI customization. He is proficient in writing Apex classes, triggers, controllers, Apex Batches, and bulk load APIs. I am also familiar with Visualforce Pages and Lighting Web Components. Read more | LinkedIn Profile
