What are Context Sources?

Context Sources define how to pull data out of Salesforce for AI analysis. Salesforce is often the central hub for customer data - even when you use other tools like Gong, Zendesk, or Gainsight, these applications typically have native connectors that sync data back into Salesforce on the Account or other objects. Features like Einstein Activity Capture also bring emails and calendar events into Salesforce.

This means Salesforce already contains a rich picture of your customer relationships. Context Sources let you tap into this data and make it available to AI for generating insights.

Examples of Context Sources:

• "Recent Tasks and Events" - Activities logged in Salesforce or synced via Einstein Activity Capture
• "Related Contacts" - Contact information linked to accounts
• "Support Cases" - Cases synced from Zendesk or Service Cloud
• "Call Recordings" - Transcripts and summaries synced from Gong or similar tools
• "Customer Health Data" - Scores and metrics synced from Gainsight

Once created, a Context Source can be reused across multiple Insight Templates and Rules. This promotes consistency and reduces maintenance effort.

How Context Sources are Structured

A Context Source has two parts:

1. The Flow - A Salesforce Flow that defines what data to retrieve. The Flow queries Salesforce objects, applies filters, and returns the relevant records.

2. The Context Source Record - A configuration record that defines:

   • The Source Object this context works with (e.g., Opportunity, Account)
   • Which Flow to execute
   • A description of the data (helps the AI understand what it's looking at)
   • The output variables returned by the Flow

Who Can Create Context Sources?

Creating Context Sources requires:

1. Precursive AI Insights Admin permission set - Grants access to create and manage Context Sources
2. Salesforce Flow Builder access - Required to create the Flows that power Context Sources

Typically, Context Sources are created by Salesforce Administrators or Operations team members with Flow building experience. Business users then select from available Context Sources when creating Rules on Insight Templates.

How Context Sources Work

Context Sources connect Insight Templates to your Salesforce data through Flows:

Insight Template → Rule → Context Source → Flow → Your Salesforce Data

When an insight is generated:

1. The system identifies which Rules are configured on the Insight Template
2. Each Rule references a Context Source
3. The Context Source's Flow is executed, passing in the record ID
4. The Flow gathers relevant data and returns it via the defined output variables
5. The AI analyzes the returned data based on the Rule's Prompt Instructions

Creating a Context Source

Step 1: Plan Your Context Source

Before creating anything, consider:

• What data do you need? (e.g., related activities, contacts, products, support cases)
• Where does the data live in Salesforce? (e.g., Task, Event, Case, custom objects from integrations)
• Which object type will this work with? (e.g., Opportunity, Account, Lead)
• Can this be reused? Design for reusability across multiple templates

Important: Review your data governance policies

Before creating Context Sources, consider your organization's policies around sharing data with AI:

• Customer data and PII should be carefully reviewed to ensure compliance with your internal policies and regulations
• Sensitive financial or contractual information may need to be excluded
• The benefit of Context Sources is that you are in full control of exactly what data is shared with the AI - nothing is sent automatically

Work with your compliance or legal team if you're unsure what data is appropriate to include.

Step 2: Create the Flow

Context Sources require a Salesforce Flow to gather data. See the section below: Building Flows for Context Sources.

Step 3: Create the Context Source Record

1. Go to the App Launcher in Salesforce
2. Search for Context Sources
3. Click New
4. Fill in the fields:

5. Click Save

Step 4: Test the Context Source

1. Create a test Insight Template for the same Source Object
2. Add a Rule that uses your new Context Source
3. Use the Preview action to generate a test insight
4. Review the Insight Result record
5. Download the context file from the Insight Result to see the exact data the Flow returned - this shows you precisely what the AI received

This verification step is important to ensure your Flow is returning the expected data before using the Context Source in production templates.

Building Flows for Context Sources

Flows are the engine behind Context Sources. They pull data from Salesforce and return it in a format the AI can analyze.

Prerequisites

To build Flows, you need:

• Salesforce Flow Builder access (typically via System Administrator profile or a custom permission set)
• Precursive AI Insights Admin permission set

Flow Requirements

For a Flow to work with Context Sources, it must meet two requirements:

1. Input Variable: recordId

Your Flow must accept an input variable that receives the ID of the record being analyzed.

This is how the system tells your Flow which record to gather data for.

2. Output Variables

Your Flow must return data through one or more output variables. You define these variables in your Flow and then list them in the Context Source's Flow Output Variables field.

Use descriptive variable names - these names appear in the context sent to the AI and help it understand what data it's looking at.

Step-by-Step: Creating a Flow

This example creates a Flow that collects recent tasks and events for an Opportunity.

Step 1: Create a New Flow

1. Go to Setup > Flows
2. Click New Flow
3. Select Autolaunched Flow (No Trigger)
4. Click Create

Step 2: Create the Input Variable

1. In the Toolbox panel, click New Resource
2. Select Variable
3. Configure:
   • API Name: recordId
   • Data Type: Text
   • Available for Input: Check this box
4. Click Done

Step 3: Get Related Tasks

1. Add a Get Records element
2. Configure:
   • Label: Get Related Tasks
   • Object: Task
   • Filter Conditions:
     • Field: WhatId
     • Operator: Equals
     • Value: {!recordId} (select your input variable)
   • Sort: ActivityDate, Descending
   • How Many Records: All Records
3. Click Done

Step 4: Get Related Events

1. Add another Get Records element
2. Configure:
   • Label: Get Related Events
   • Object: Event
   • Filter Conditions:
     • Field: WhatId
     • Operator: Equals
     • Value: {!recordId}
   • Sort: ActivityDateTime, Descending
   • How Many Records: All Records
3. Click Done

Step 5: Create the Output Variables

Create separate output variables with descriptive names:

For Tasks:

1. In the Toolbox panel, click New Resource
2. Select Variable
3. Configure:
   • API Name: recentTasks
   • Data Type: Record
   • Allow Multiple Values (Collection): Check this box
   • Object: Task
   • Available for Output: Check this box
4. Click Done

For Events:

1. Click New Resource again
2. Select Variable
3. Configure:
   • API Name: recentEvents
   • Data Type: Record
   • Allow Multiple Values (Collection): Check this box
   • Object: Event
   • Available for Output: Check this box
4. Click Done

Step 6: Assign Results to Output Variables

1. Add an Assignment element

2. Configure:

   • Label: Assign Tasks
   • Variable: {!recentTasks}
   • Operator: Equals
   • Value: {!Get_Related_Tasks}

3. Click Done

4. Add another Assignment element

5. Configure:

   • Label: Assign Events
   • Variable: {!recentEvents}
   • Operator: Equals
   • Value: {!Get_Related_Events}

6. Click Done

Step 7: Connect and Save

1. Connect the elements in order: Start > Get Tasks > Get Events > Assign Tasks > Assign Events
2. Click Save
3. Give your Flow a name (e.g., "Gather Opportunity Activities")
4. Note the API Name - you'll need this for the Context Source
5. Click Activate

Step 8: Configure the Context Source

When creating the Context Source record, set the Flow Output Variables field to:

recentTasks,recentEvents

Tips for Effective Context Sources

Design for Reusability

Create Context Sources that can be used across multiple Insight Templates:

• Use generic names like "Recent Activities" rather than "Sales Team Activities"
• Avoid hard-coding specific criteria that might not apply everywhere

Keep Data Relevant

Only gather data that will help the AI provide useful insights. Too much irrelevant data can dilute the quality of insights.

Good: Recent activities from the last 90 days
Less effective: All activities since the record was created

Use Filters Wisely

Add filter conditions to focus on meaningful data:

• Filter by date ranges (e.g., last 30 days)
• Filter by status (e.g., only open cases)
• Filter by type (e.g., only customer-facing activities)

Consider Performance

Large data sets can slow processing and increase costs. Consider:

• Using sort and limit settings to get the most recent/relevant records
• Filtering out low-value data
• Adding date range filters

Document Your Context Sources Well

Use the Context Description field to clearly describe:

• What data this Context Source gathers
• Where the data comes from (e.g., "Emails synced via Einstein Activity Capture", "Cases from Zendesk integration")
• Any filters or date ranges applied

This description helps both you and the AI understand what data is being analyzed. The Rule's Prompt Instructions will then define how the AI should interpret and use this data.

Troubleshooting

"Flow not found" Error

• Verify the Flow API Name is spelled correctly (case-sensitive)
• Ensure the Flow is Activated
• Check that the Flow type is Autolaunched (not Screen Flow)

No Data Returned

• Test the Flow in Debug mode with a valid record ID
• Verify your filter conditions aren't too restrictive
• Check that related records actually exist for your test record
• Download the context file from the Insight Result to see exactly what was returned

Variable Name Errors

• Input variable must be named exactly recordId
• Output variable names must match what's listed in the Context Source's Flow Output Variables field
• All variables must have the correct "Available for Input/Output" settings checked

Timeout or Performance Issues

• Reduce the amount of data being gathered
• Add date filters to limit to recent records
• Use record limits on Get Records elements