Getting Started with Salesforce Agent Setup
Overview
This Salesforce setup guidance is intended as a high-level orientation and starting point only. The official source of product documentation, setup steps, and requirements for any referenced features is Salesforce, and those materials may change at any time. You should always rely on Salesforce’s current help articles and Trailhead content for detailed, current instructions, and use this guide simply to review key prerequisites and point to relevant resources rather than as a replacement for official Salesforce documentation.
Before You Begin
Complete the Salesforce platform setup items in this section before configuring agents. These prerequisites are the items you should confirm, enable, or request from Salesforce before beginning the topic, action, and flow build described in other agent implementation guides.
Required Enablement Resources
Use these Salesforce resources to prepare your Org:
- Get Ready for Agentforce (Overview trail)
- Explore Agentforce for Service: Key Benefits & Setup
- Agent Planning
- Set Up Your Knowledge Base (Lightning Knowledge Basics)
- Optimize Salesforce Knowledge Setup
- Utilize Agentforce Knowledge for Enhanced Service
- Salesforce Knowledge Data Categories Best Practices and Considerations
- Use Salesforce Knowledge for Agentforce Data Libraries
- Implementing Data Cloud for Enhanced Agentforce
- Explore Data Cloud and Agentforce
- Enable Digital Experiences (Trailhead)
- Digital Experiences
- Connect an Agent to Messaging (Agentforce)
- Connect Your AI Agent to Customer Channels (Agentforce Messaging/Web)
- Configure Messaging for Web Deployment (Messaging for In App & Web)
- Service Cloud Voice Implementation Guide
- Improve Customer Service with Service Cloud Voice
- Set Up Service Cloud Voice with Partner Telephony
- New Agentforce Builder UI
Salesforce-requested enablement items
In addition to standard setup tasks, some capabilities may require you to request activation, provisioning, or enablement from Salesforce before your institution’s implementation team can fully use them.
Refer to the following table during your Project Kickoff:
|
Capability |
What to confirm or request |
Related Salesforce Resource |
|---|---|---|
| Agentforce access | Confirm Agentforce is provisioned in the environment and that implementation users have access to build and manage agents. | Explore Agentforce for Service: Key Benefits & Setup |
| Messaging for In-App and Web | Confirm Messaging is enabled and that the org can create messaging channels and deployments. | Connect an Agent to Messaging |
| Knowledge for agent answers | Confirm Salesforce Knowledge is enabled, and that article access, layouts, and categories are available for setup. | Set Up Salesforce Knowledge |
| Agent grounding with knowledge/files | Confirm whether the client will use an Agentforce Data Library and whether prerequisites for knowledge-based grounding are in place. | Utilize Agentforce Knowledge for Enhanced Service |
| Data 360/Data Cloud | Confirm whether Data 360 or Data Cloud must be activated. Salesforce notes Data 360 needs to be activated before setting up a Data Library. | Implementing Data Cloud for Enhanced Agentforce |
| Digital Experiences | Confirm Digital Experiences is enabled if the client will deploy messaging in an Experience Cloud site. | Enable Digital Experiences and Set Up Messaging |
| Service Cloud Voice | Confirm whether voice is in scope and whether Service Cloud Voice setup and any related provisioning need to be completed before voice deployment. | Service Cloud Voice Implementation Guide |
High-level Prerequisite Checklist
Before configuring the FAQ agent, confirm the following:
Licensing
Confirm Agentforce is available in the org, and your implementation team has the required access to Agentforce Builder and related setup areas.
- Licensing & Edition:
- Supported editions: Enterprise, Unlimited, and Performance
- Agentforce / Einstein AI add-on licenses (varies by agent type)
Enable
You must enable:
Lightning Experience (not Classic)
Einstein Generative AI turned on (Agentforce cannot exist without Einstein — it’s the AI runtime and trust layer).
Once available, you must toggle on ‘Einstein’.
Einstein Copilot/Agent Studio (Optional but often needed)
Agentforce
Once available, you must toggle on ‘Enable Agentforce’. After this, you should see the ‘New Agent button.’
Data Cloud (recommended for better data grounding). Data enablement for Agentforce is reviewed and completed, including Data 360 / Data Cloud activation if the client plans to use an Agentforce Data Library or other grounded data features.
Salesforce Knowledge is enabled, article access is configured, data categories are configured as needed, and approved knowledge content is published.
Messaging is enabled in Salesforce, and the org is prepared for Messaging for In-App and Web.
A messaging channel, routing configuration, queue, and deployment are created for the web experience where the agent will appear.
If the deployment will only show during staffed hours, business hours are configured and associated with the deployment behavior as needed.
If voice is in scope, Service Cloud Voice with Amazon Connect or another supported telephony approach is selected and prepared before voice-specific agent work begins.
Security & Permissions
Note: adjust these permissions to align with your institution’s requirements. The list below represents the minimum permissions needed.
Proper profiles/permission sets for users building or testing the Agent:
“Manage AI Agents” permission required.
Access to Agentforce Studio.
Access to Einstein Generative AI.
Access to edit or run flows.
Access to Object/Fields the agent will touch.
Data access controls for the Agent user because Agents act on your data, so access control defines what they can “see” and do:
Assess Role hierarchy via Roles and Org-Wide Defaults.
Agentforce inherits the role hierarchy of the user interacting with it and only queries what the user can see.
Object/field-level security for any object the agent will need to affect.
Access to run flows/actions.
Access to Knowledge articles and data categories.
Data Readiness:
Clean, structured CRM data (AI outputs are only as good as the underlying data)
Standardized objects (Contact, Cases, Task, Inquiry, etc.)
Knowledge articles (for service agents)
Defined automations (Flows, etc.)
Summary
Agentforce is not a simple feature you can switch on—it relies on a well-prepared Salesforce foundation. At a high level, this includes the correct licenses and feature enablement (Agentforce and Einstein Generative AI), a properly configured platform (Lightning Experience with Einstein enabled), a defined security model (permissions, role hierarchy, and MFA), and clean, structured data the AI can trust. Equally important is defining clear use cases and the actions the agent will perform before you begin building. These prerequisites are critical because Agentforce operates entirely within your organization’s data, permissions, and automation. Without this foundation, the agent may produce inaccurate results, fail to complete actions, or expose inappropriate information, increasing both risk and rework. Proper preparation ensures your agent is accurate, secure, and effective.