An Agentforce at Universal Containers (UC) is building with no-code tools only. They have many small accounts that are only touched periodically by a specialized sales team, and UC wants to maximize the sales operations team's time. UC wants to help prep the sales team for the calls by summarizing past purchases, interests in products shown by the Contact captured via Data Cloud, and a recap of past email and phone conversations for which there are transcripts.
Which approach should the Agentforce Specialist recommend to achieve this use case?
A. Use a prompt template grounded on CRH and Data Cloud data using standard foundation model.
B. Fine-Tune the standard foundational model due to the complexity of the data
C. Deploy UC's own custom foundational model on this data first.
A. Use a prompt template grounded on CRH and Data Cloud data using standard foundation model.
Explanation:
Universal Containers (UC) wants to use no-code tools to help prep the sales team by summarizing:
1. Past purchases (likely stored in CRM Analytics or standard Salesforce objects)
2. Product interest data captured in Data Cloud
3. Recaps of email and call transcripts (stored or accessible via CRM objects or Einstein Conversation Insights)
The most scalable and no-code-friendly approach is to:
✅ Use a Prompt Template in Prompt Builder, grounded on:
CRM (Salesforce record data)
Data Cloud (for unified customer profiles and engagement)
and optionally CRM Analytics (CRMA) or Einstein Conversation Insights
This lets Einstein Copilot or Agentforce automatically:
Summarize structured + unstructured data
Use Salesforce-native LLMs (standard foundation models) without requiring fine-tuning
Avoid custom model hosting or code-based solutions
❌ Why the other options are incorrect:
B. Fine-tune the standard foundational model
Fine-tuning is not supported on Salesforce-managed LLMs.
It also requires code/data science resources, and goes against the “no-code” requirement.
C. Deploy UC's own custom foundational model
This is heavy, costly, and unnecessary.
Requires Einstein Studio, MLOps setup, and custom integration — which does not align with the no-code approach UC wants.
🔗 References:
Salesforce Help — Prompt Builder Overview
Salesforce Docs — Data Cloud Grounding for Prompt Templates
Salesforce Einstein Studio vs Prompt Builder