Prompt Builder
Prompt Builder is Salesforce's tool for designing reusable prompt templates that combine generative AI with Salesforce data.
Definition
Prompt Builder is Salesforce's tool for designing reusable prompt templates that combine generative AI with Salesforce data. Each template defines a prompt for a large language model, with merge fields that pull live data from Salesforce records, related records, Knowledge articles, Data Cloud objects, and other grounded sources. The platform substitutes the merge fields at runtime, sends the assembled prompt to an LLM, and returns the generated text. Prompt templates power email generation, sales summaries, service response drafting, and content creation throughout the Salesforce UI and across Agentforce agents.
Prompt Builder is part of Einstein's generative AI stack. A prompt template has a type (Sales Email, Field Generation, Flex, Record Summary), an LLM target (Salesforce-hosted, OpenAI via Trust Layer, Anthropic Claude, or custom), grounding sources that pull contextual data, merge fields that inject record values, and a system prompt that frames the LLM's task. Admins build templates without writing code; developers can extend with Apex for advanced grounding. Templates surface in record pages, Sales Engagement, Service Console, Marketing Cloud, and as actions inside Agentforce.
How Prompt Builder makes generative AI useful inside Salesforce
Template types and their use cases
Prompt Builder supports several template types, each tuned for a specific use case. Sales Email generates personalized outbound emails grounded in Account and Opportunity data. Field Generation produces values for specific fields (a summary description, a recommended next step). Record Summary creates narrative summaries of records and their related data. Flex is the generic template type for any other prompt-driven content generation. Each type has its own input parameters and output format expectations.
Merge fields and the grounding model
The defining feature of Prompt Builder is grounded merge fields. Where traditional email templates use simple {!Account.Name} merge syntax, Prompt Builder pulls richer contextual data: full Account record, related Opportunities, recent Activities, matching Knowledge articles, Data Cloud unified profiles. The platform fetches this data at runtime, substitutes it into the prompt, and sends the assembled context to the LLM. Strong grounding is the reason Salesforce-generated content is more accurate than generic ChatGPT output.
LLM selection and the Trust Layer
Each template specifies which LLM to use. Salesforce hosts its own models (xGen, the Salesforce-trained foundation models). The Einstein Trust Layer connects to third-party LLMs (OpenAI, Anthropic Claude, Google Gemini, AWS Bedrock) with zero-data-retention agreements that protect Salesforce data from being used for model training. Customers can also bring their own LLM endpoints. The Trust Layer applies data masking, toxicity filtering, and audit logging regardless of which model the template uses.
Prompt iteration and the Prompt Studio
Prompt Studio is the interactive editor where admins build and test templates. The left panel shows the prompt definition; the right panel shows preview results against sample data. Edit the prompt, click Preview, see the generated output. Iterate until the output meets quality bars. The Studio also shows the resolved prompt (after merge field substitution) so admins can debug why the LLM produced unexpected results. Strong prompts come from many iterations in the Studio, not from one-shot writing.
Activation: where templates appear
Each template can be activated for specific surfaces. Sales Email templates appear in Sales Engagement and inline email drafting. Field Generation templates appear on record pages with the Einstein component. Record Summary templates power AI summary panels. Flex templates expose as actions in Agentforce and Flow. The activation choice determines where users see the generated content; mismatched activation hides good templates from their intended users.
Apex integration for advanced grounding
For grounding that exceeds what the declarative merge field picker supports, developers can write Apex methods marked with the appropriate annotations to inject custom data into the prompt. The Apex method runs at template invocation, returns structured data, and the platform substitutes it into the prompt. This is the path for templates that need cross-org data, custom calculations, or external system integration. Most templates do not need Apex; complex ones with bespoke grounding requirements do.
Monitoring, evaluation, and credit consumption
Each template invocation consumes Einstein Credits proportional to the LLM tokens used. Setup > Einstein Generative AI > Usage shows consumption per template. Build monitoring around credit spend so high-volume templates do not produce surprise overages. Salesforce also surfaces quality signals (user thumbs up/down on generated content) that admins should review to identify templates needing iteration.
How to build a Prompt Template
Building a prompt template is iterative. Define the use case, write the initial prompt, ground with the right data, preview against realistic records, refine the wording until output quality is consistent. Most of the value is in iteration; one-shot prompts rarely produce production-quality results.
- Define the use case and audience
What content should the template generate? Who consumes it? Sales outbound emails for prospects, service response drafts for agents, summaries for executives. Clarity on use case drives every downstream choice.
- Open Prompt Builder and create a new template
Setup > Prompt Builder > New Prompt Template. Pick the template type (Sales Email, Field Generation, Record Summary, Flex). Provide a name and description.
- Pick the LLM target
Choose the LLM. Salesforce-hosted models for default use. OpenAI, Anthropic, or other Trust Layer models for specific quality or feature needs. Bring-your-own LLM for custom endpoints.
- Configure grounding sources
Pick the records and related data the prompt needs. Current record, related records, Knowledge, Data Cloud objects, custom Apex data. Strong grounding is the difference between accurate output and hallucinated content.
- Write the prompt with merge fields
Draft the prompt text. Insert merge fields from the grounding sources where appropriate. Frame the LLM''s task clearly: be specific about tone, length, format. Vague prompts produce inconsistent output.
- Preview against sample records
Use the Preview panel to test against real records. Review the resolved prompt (after merge field substitution) and the LLM''s response. Iterate on the prompt wording, grounding, and merge fields until output is consistent and accurate.
- Activate the template for the right surfaces
Configure where the template appears: Sales Engagement, record page Einstein component, Agentforce action, Flow. The activation makes the template available to users; without it, the template stays admin-only.
- Monitor usage, quality, and credit consumption
Setup > Einstein Generative AI > Usage. Track which templates run most often, which produce thumbs-up versus thumbs-down feedback, and the credit consumption. Iterate on weak templates and budget for high-use ones.
Sales Email, Field Generation, Record Summary, or Flex. Drives the input parameters and expected output format.
Salesforce-hosted, Trust Layer third-party (OpenAI, Anthropic), or bring-your-own. Different cost and quality trade-offs.
Records, related data, Knowledge, Data Cloud, or custom Apex. Determines what real data the prompt draws on.
- Ungrounded prompts hallucinate. The LLM confidently states facts that are not in your data unless you ground the prompt with real Salesforce context. Always configure grounding for data-specific output.
- Prompt quality requires iteration. One-shot prompts rarely produce production-quality output. Plan multiple Preview cycles and refine the wording until results are consistent.
- Einstein Credits consume per invocation. High-volume templates produce significant credit spend. Monitor consumption and plan budget accordingly.
- Vague prompts produce inconsistent output. Be specific about tone, length, format, and what to do when data is missing. The LLM follows instructions; ambiguous instructions produce variable output.
- Template activation gates where users see the output. Forgetting to activate for the right surface means the template stays admin-only and produces no business value.
Trust & references
Cross-checked against the following references.
- Prompt Builder OverviewSalesforce Help
- Create a Prompt TemplateSalesforce Help
Straight from the source - Salesforce's reference material on Prompt Builder.
- Prompt Builder DocumentationSalesforce Help
- Grounding PromptsSalesforce Help
- Einstein Trust LayerSalesforce Help
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What is Prompt Builder?
Q2. What's 'grounding' in this context?
Q3. Why is template design important?
Discussion
Loading discussion…