In the fast-moving world of marketing, segmentation has always been the bedrock of personalized campaigns. But let’s face it: building audience segments has traditionally been a manual, time-consuming process. From sorting through data extensions to writing complex queries, marketers often spend hours just preparing to launch a single campaign.
Now imagine this: instead of building segments manually, you simply ask. With Salesforce’s latest AI innovations—Einstein Prompt Builder, AgentForce, and MCP (Model Context Protocol)—that’s exactly the future we’re stepping into.
You Used to Build Segments. Now, You Just Ask.
Whether you’re targeting high-intent buyers or re-engaging at-risk customers, building segments used to mean navigating data fields, crafting SQL queries, and hoping it all aligned with your campaign goals. But with Salesforce AI, that entire workflow transforms into a single natural language prompt.
For example:
“Generate three high-conversion segments for the monsoon skincare campaign using email engagement and last 60-day purchases.” In seconds, AI delivers three dynamic, ready-to-deploy segments—grounded in your actual data. No code. No guesswork. Just outcomes.
Common Use Cases: Now Automated
Here are a few types of segments marketers typically build (manually):
- High-Intent Buyers: Recent site visits + cart adds + past purchases
At-Risk Customers: Low engagement over 90 days + high return rate
Loyal Customers: Repeat buyers + NPS ≥ 9 + high average order value
Campaign-Responsive Users: High open/click rates in recent campaigns
Seasonal Shoppers: Only active during festivals or month-end sales
With AI, these are now accessible in seconds—just by describing what you need.
The Intelligence Behind the Simplicity
So how does this prompt-to-segment magic happen? Here’s a breakdown of the intelligent architecture working behind the scenes:
- Prompt Initiation: Your prompt is entered via the Marketing Cloud Advanced UI and routed to Einstein Prompt Builder.
- Einstein Prompt Builder
This module:
– Classifies your request (e.g., segmentation vs. personalization)
– Interprets your intent
– Adds formatting rules
– Passes the enhanced prompt to AgentForce for context-based processing - AgentForce: AgentForce identifies the task type—such as “personalization using historical behavioral data”—and invokes MCP to fetch real-time, org-specific metadata:
– Data extensions and field mappings
– Subscriber keys and permissions
– Custom fields like skin type, loyalty tier, or engagement history - MCP (Model Context Protocol)
This is where the contextual magic happens. MCP:
– Pulls schema and segmentation logic directly from your Salesforce environment
– Honors field-level security and user roles
– Retrieves active segment templates and configurations
– Packages all this into a structured Context Object, which it passes back to the AI engine
From Prompt to Output: AI That Understands Your Ecosystem
Once the AI (LLM) receives both your original prompt and the metadata context, it:
The result? AI-generated segments that actually work with your data model—no surprises, no debugging.
Example output:
Segment A: High Open Rate, High Spend
Segment B: Low Engagement, Interested in New Products
Segment C: Recent Purchasers, Loyalty Tier 1
It may even suggest journey triggers or content variants to help you take immediate action.
Seamless Integration with Marketing Cloud Tools
Once segments are generated, they can be deployed directly into:
What used to be a disconnected, manual process is now a cohesive, AI-powered workflow.
From Rule-Based to Intent-Driven Marketing
This isn’t just AI creating content—it’s creating strategy.
With Einstein Prompt Builder, AgentForce, and MCP segmentation becomes:
Ready to Work Smarter?
If you’re ready to stop building segments and start driving outcomes, it’s time to rethink how your team uses AI.
Let Lirik help you put AI to work—so you can put your time where it matters most.
