How to Create a record summary Prompt Template in Salesforce?

To create a Record Summary Prompt Template in Salesforce, use Prompt Builder to write a short summary of a record (like an Account or Opportunity) using AI. It helps users quickly understand key details without reading every field.

Here’s a simple step-by-step guide:

Steps to Create a Record Summary Prompt Template

Open Prompt Builder

In Setup, search for Prompt Builder.

Click Prompt Builder → Then click New Prompt Template.

Choose Template Type

Select Record Summary as the template type.

Enter Basic Info

Give your template a Name and Description.

Example:

Name: Case Record Summary Template

Description: This template is used to summarize case record details

In the next step, add prompt and select one existing record to preview and finally activate the template.

Now, let's design a screen flow as shown below.

Image 1:

Image 2: 

Image 3:

Case Summary Screen:

Display Element inside Case Summary Screen:

To use this flow  on a Lightning record page, add this component to lightning record page under Lightning App Builder for Case object as shown below.

Now, it is time to see the changes in action, open case and view the results as shown below.

How to Create a Field Generation Prompt Template in Salesforce?

Salesforce now lets you use generative AI to automatically fill in fields on Lightning record pages. This feature is called a Field Generation Prompt Template, and it helps users quickly add summaries or descriptions using AI. It’s a great way to make customer conversations more productive and save time on manual writing.

Let’s walk through how to set it up in simple steps.

Steps to Create a Field Generation Prompt Template

Go to Prompt Builder

In Setup, search for Prompt Builder in the Quick Find box.

Click on Prompt Builder.

Click New Prompt Template.

In the dropdown, choose Field Generation as the template type.

Fill in Basic Details

Give your template a name.

Choose the Object that contains the field you want AI to fill (like Account, Case, etc.).

In the next step, add prompt and select one existing record to preview and finally activate the template.

To use this template on a description field on a Lightning record page, Dynamic Forms must be enabled. Enable Dynamic Forms for case object in lightning record page under Lightning App Builder.

Note: Not all objects support Dynamic Forms. Check Salesforce’s list of supported objects.

Now, Choose the Field "Description" in our case and select the prompt template we created just now as shown below.

Save and Activate the lightning record form. 

Now, open one case and you will be able to see the icon   appears on description field as shown below. If you click on icon you will be able to see Einstein in action as shown in second image and the text generated can be used after clicking use button. The text will be added to description field automatically and can be save on case.

What Is Salesforce Prompt Builder and How to Enable It?

Salesforce is rapidly evolving with AI-powered features, and one of the most exciting additions is Prompt Builder. Designed to empower admins and developers, Prompt Builder turns you into a prompt engineer without needing to write code. But what exactly is it, and how do you enable it in your org? Let’s dive in.

What Is Salesforce Prompt Builder?

Prompt Builder is one of its most powerful tools. It allows admins and developers to create intelligent, data-driven prompts that communicate directly with large language models (LLMs) like ChatGPT. Whether you're automating service responses or generating sales emails, Prompt Builder puts generative AI at your fingertips.

Why It Matters?

No-code AI integration: Admins can build prompts without writing code.

Data-aware outputs: Prompts can include merge fields from Salesforce records.

Reusable templates: Create once, use across flows, Apex, or screen components.

Preview and test: Simulate outputs before deploying to users.

What Is a Prompt Template in Salesforce?

A Prompt Template in Salesforce is a reusable, structured instruction that guides a large language model (LLM) to generate intelligent, context-aware responses using real-time Salesforce data.

How It Works?

Prompt Templates use merge fields to dynamically pull data from Salesforce records. These fields are written in a special syntax like:

{!$Input:Sender.Name}

{!$Input:Recipient.MailingCity}

Here is one example:

Case Summary for Support Agent

Case {!$Input:Case.CaseNumber} was opened by {!$Input:Contact.Name} regarding {!$Input:Case.Subject}. The customer is located in {!$Input:Contact.MailingCity} and uses {!$Input:Product.Name}. Previous interactions include {!$Input:Case.PreviousComments}. Recommend next steps based on similar resolved cases.

When this runs, Salesforce fills in the blanks with actual data—so the AI knows exactly who’s speaking, who they’re speaking to, and what context to use.

How to Get Started with Salesforce Prompt Builder?

Before you can start building intelligent prompt templates, you need to enable the right Einstein features and assign permissions. Here's a quick 3-step guide to get you up and running:

Turn On Einstein Setup

Go to Setup and search for Einstein Setup. This activates the foundational AI capabilities across your org.

Enable Einstein for Sales

Still in Setup, search for Einstein for Sales and enable it. This unlocks generative AI features like Prompt Builder, Sales Emails, and AI-powered recommendations.

Assign Permission Sets

To access and use Prompt Builder, assign the following permission sets:

Prompt Template Manager – for users who create and manage templates

Prompt Template User – for users who run the templates

You can assign these via:

Setup → Permission Sets → Manage Assignments

Once these steps are complete, you’ll see Prompt Builder available in Setup—and you’re ready to start designing AI-powered templates!

What Is a Prompt—and Why Writing It Right Matters

In the age of AI, the word “prompt” has taken on a whole new level of importance. Whether you're chatting with an AI assistant, generating images, or building automation workflows, prompts are the bridge between your intent and the machine’s output. But what exactly is a prompt—and why does writing it the right way make all the difference?

What Is a Prompt?

A prompt is a piece of input or instruction given to an AI system to guide its response. Think of it as a question, command, or description that tells the AI what you want. It can be as simple as:

“Write a poem about the moon.”

Or as complex as:

“Generate a 500-word blog post explaining the importance of cybersecurity for small businesses, using a friendly tone and real-world examples.”

In essence, a prompt is your way of communicating with AI—your words become the blueprint for its output.

Why Writing the Right Prompt Is Crucial?

Crafting a good prompt isn’t just about typing something and hoping for the best. It’s about being intentional, clear, and strategic. Here’s why it matters:

Precision Drives Quality

A vague prompt leads to vague results. If you say “Tell me about marketing,” the AI might give you a generic overview. But if you say “Explain digital marketing strategies for B2B startups in 2025,” you’ll get a focused, relevant answer.

Context Shapes Creativity

AI thrives on context. The more details you provide—tone, audience, format, purpose—the better the output. For example:

“Create a formal email to a client about project delay” vs.

“Write a casual update to a teammate about running late on a task”

Same topic, different tone—prompting makes the difference.

Saves Time and Reduces Rework

A well-written prompt minimizes back-and-forth. Instead of refining outputs repeatedly, you get closer to your goal on the first try. This is especially important in professional settings where time is money.

Unlocks Advanced Capabilities

Want to generate code, design a logo, or simulate a business scenario? The right prompt can unlock powerful features. For example:

“Generate Python code to scrape product prices from an e-commerce site and store them in a CSV.”

That’s not just a request—it’s a command that activates a specific skill set.

Tips for Writing Better Prompts

Here’s how to level up your prompting game:

Be specific: Include keywords, constraints, and goals.

Define the format: Blog post, list, email, poem—tell the AI what structure you want.

Set the tone: Formal, friendly, persuasive, humorous—tone matters.

Clarify the audience: Is it for beginners, experts, clients, or students?

Use examples: If possible, show what you mean.

Final Thoughts

Prompts are not just inputs—they’re instructions, inspirations, and invitations. Writing them well is a skill that empowers you to get the most out of AI tools. Whether you're a developer, marketer, educator, or just curious, mastering the art of prompting is your gateway to smarter, faster, and more creative results.

How Salesforce Agents Think, Act, and Get Things Done

Salesforce Agentforce is a powerful AI tool that helps automate tasks, answer questions, and guide users—all with minimal human help. But how does it actually work? And what are its key parts? Let’s explore how Agentforce agents are built and how they operate.

The Core Building Blocks of an Agent

Every Agentforce agent is made up of five key components:

Agent: The AI assistant itself. It’s designed to reduce workload and boost productivity by handling routine and complex tasks. Agents can work independently, spot opportunities, and take action within the rules you set.

Topics: These are categories of tasks the agent can handle. For example, a topic like “Deal Management” might include actions to find contacts, log calls, or create to-do items.

Actions: These are the steps the agent takes to complete a task. Actions can include sending messages, updating records, or drafting emails using Salesforce data. You can use built-in actions or create custom ones for your business.

Reasoning Engine: This is the brain that helps the agent decide what to do. It reads the user’s message or trigger, picks the right topic, and plans how to reach the goal using available actions.

Large Language Model (LLM): This is the smart technology that helps the agent understand language and respond naturally. The reasoning engine calls the LLM multiple times during a task to make sure the agent understands and reacts correctly.

How Agentforce Works Step-by-Step

Let’s follow the journey of a user message through the Agentforce system:

1. A Message or Trigger Starts It All

The process begins when a user sends a message or when a trigger (like a lead assignment or data change) activates the agent.

2. The Agent Picks a Topic

The agent compares the message to all available topics and picks the one that best matches the task. If no match is found, it switches to an “Off-Topic” mode and tries to guide the user back.

3. The Agent Starts Reasoning

The agent uses the reasoning engine to:

Run actions like flows or Apex classes

Ask for more details or clarifying questions

Respond directly if it already has the answer

It follows topic instructions carefully—for example, asking for a model number before giving product info or verifying identity before sharing account details.

The agent loops through this reasoning process up to seven times to make sure it gets things right.

4. Final Response Check (For Service Agents)

Before replying, the agent checks if the answer is safe, accurate, and based on trusted data. If the response doesn’t pass, it creates a new one. If it still can’t help, it lets the user know.

5. The Agent Sends the Reply

Finally, the agent sends the message. The user can reply or ask something new, and the agent continues the conversation smoothly.