Test your way to smarter Service Replies

Test Your Way to Smarter Service Replies Using Prompt Builder

By

If your service reps are feeling the pressure, they’re not alone. According to the latest State of Service report, 86% say customer expectations are higher than ever. That’s where you come in.

As an admin, you’re the one making sure your team has the tools to meet those expectations. Enter: Service Replies. This artificial intelligence (AI)-powered feature suggests responses—right in the service console—while a chat is happening. The magic? It can pull from both the conversation history (contextual) and your company’s knowledge base (grounded) to offer accurate, helpful replies.

And now, it gets better. You’re no longer stuck with out-of-the-box responses that feel a little… canned. With Service Replies integrated into Prompt Builder, you can customize the templates to reflect your business tone, processes, and customer needs. Think brand-consistent, on-message replies—at scale.

But before you hit activate, you’ll want to make sure everything works the way you expect. That’s where in-app testing comes in. Prompt Builder lets you preview and test your prompts directly, so you can tweak the tone, fix glitches, and make sure your reps are always set up for success.

Your reps need answers that sound human and stay on-brand. That means your prompts need to be dialed in before you hit activate. Luckily, Prompt Builder gives you a sandbox for just that.

Test Contextual prompt templates

Contextual templates are the first step. These generate replies based on the chat history. Here’s how to test:

  1. Go to Setup and search for “Prompt Builder”.
  2. Choose either:
    • Draft Service Replies from Context – Live Chat Transcript, or
    • Draft Service Replies from Context – Messaging Session
  3. On the Test Inputs tab:
    • Paste a conversation into the Transcript field.
    • Choose a conversation object (required for preview).
    • Click Preview to generate a reply.

Prompt Builder interface showing the test inputs, resolution, and response generated based on a Contextual Service Replies prompt template.

Now read it like your customer would. Does it sound like your brand? Is it helpful and clear? If not, adjust the prompt. Use Save As → New Version to iterate.

Test Grounded prompt templates

Grounded templates raise the stakes by pulling in knowledge from your company’s data. More accurate? Yes. A little trickier to set up? Also yes. In order to set up grounding for Service Replies, you’ll need to first create an Agentforce Data Library and assign it to the Service Replies feature in setup. Once that’s done, Service Replies will be able to retrieve relevant passages from your knowledge base via Data Cloud and use them to hydrate your prompt template. Now you’re ready to test.

Here’s the process:

  1. In Prompt Builder, pick either:
    • Draft Service Replies with Grounding – Live Chat Transcript, or
    • Draft Service Replies with Grounding – Messaging Session
  2. Open the Test Inputs tab and:
    • Paste a chat transcript.
    • Choose a conversation object.
    • For best results, copy a search_query from a Contextual template’s preview output and paste it into the LLM Search Query field.
    • Find your Retriever ID (from Einstein Studio > Retrievers) and add it.

Click Preview, then check: Is the response accurate? Is it pulling from the right source? Make adjustments just like you would with Contextual prompt templates.

Pro tip: Use consistent tone and style instructions across both templates to keep the experience seamless for your customers.

Prompt Builder interface showing the test inputs, resolution, and response generated based on a Grounded Service Replies prompt template.

Tips for testing and refinement

Testing isn’t one-and-done. Use real-world examples that match your service scenarios: product FAQs, order tracking, troubleshooting, etc. Try a mix of:

  • Short versus long conversations
  • Different stages of a service interaction
  • Common and edge-case questions

Watch for prompt templates that fall flat in certain scenarios and tweak accordingly. Prompt tuning is ongoing—but that tenacity is what makes you a great admin.

Keep refining. Keep testing. And make every reply count.

Resources

100 Ready-to-Use AI Prompts for Salesforce Admins

100 Ready-to-Use AI Prompts for Salesforce Admins

In this blog post, we looked at how the writing skills you already use every day carry over to prompting an agent. Now let’s put them to work even more. At our Dreamforce 2026 session on Agentforce Vibes, Lay Durafe and I handed out something to practice on: 100 ready-made prompts for the jobs admins […]

READ MORE
8 Lessons for Writing AI Prompts as a Salesforce Admin

8 Lessons for Writing AI Prompts as a Salesforce Admin

Agents are turning into everyday assistants for your work, and AIforce gives those assistants a direct line into your Salesforce org. At Dreamforce 2026, my colleague Lay Durafe and I showed how Salesforce admins can use the same tools developers use to vibe code, and vibe admin instead. The more you work with agents, the […]

READ MORE
introduction to AIforce in Action

Introduction to AIforce for Admins

In September, Salesforce announced AIforce, its strategy for bringing Salesforce to any AI model, agent, or interface where your teams already work. To see where it fits, picture the Salesforce platform as four layers. Data 360 brings your data together. Customer 360 holds your workflows, data models, and business logic. Agentforce is where agents get […]

READ MORE