In this Blog Post, you will learn how to configure Salesforce Agentforce Coworker. It also covers architecture, Data 360 integration, licensing models, and step-by-step setup for technical teams.
Salesforce Agentforce Coworker: Architecture, Pricing, and Step-by-Step Setup Guide
As enterprise environments grow more complex, technical teams often find themselves context-switching across isolated applications, legacy databases, and disparate support channels. Salesforce Agentforce Coworker directly addresses this challenge by providing an autonomous, conversational interface that bridges your entire work ecosystem into a unified workspace.
Powered by Data 360, Agentforce Coworker allows developers, architects, and administrators to surface intelligence, automate tasks, and streamline cross-team workflows without leaving Salesforce.
Technical Overview & Architecture
Agentforce Coworker acts as a centralized intelligence layer across your org and connected platforms. By leveraging Data 360, it grounds AI interactions in real-time enterprise data, ensuring governance, accuracy, and operational efficiency.
Key Capabilities
- Unified Workspace: Consolidates tools, data streams, and communication channels into one conversational interface.
- Cross-System Connectivity: Integrates external repositories—such as Google Drive and Microsoft SharePoint—by connecting them directly to Data 360.
- Zero Context-Switching: Enables users to query data, trigger flows, and interact with customers from a single entry point.
Licensing & Pricing Models
Salesforce offers two flexible models to align with enterprise consumption requirements:
- Seat-Based Pricing: Available through Agentforce 1 Edition or dedicated Agentforce for Sales, Service, and Industries licenses.
- Usage-Based Pricing: Managed through Flex Credits or Data Services Credits, ideal for variable volume workloads.
Setup & Configuration Guide
Follow this step-by-step guide to activate and deploy Agentforce Coworker across your org.
Prerequisite
Ensure you have assigned the Agentforce Coworker Admin Permission Set to your administrator account before proceeding.
Implementation Steps
Steps:
1. Go to Get Started with Agentforce Coworker in Salesforce Setup.
2. Click Get Started.
3. In "Turn on Agentforce Coworder", toggle on.
4. Wait for the Progress Indicator to complete.
5. All the steps should be completed to proceed further.
6. In "Manage Data" section, enable the Data Sources.
7. In the Turn on End User Experience (Optional) section, click the Manage link.
8. Read and check the consent box to confirm you have read and agreed to the terms.
9. Click the Turn On button.
10. In the Manage Users section, click the Manage link.
11. In the Assign Permission Groups to Users window , select the users to grant access to the Coworker.
12. Click the Assign button.
13. Check the Assignment Summary.
14. Click the Done button.
15. To test open an App and use the Ask button next to the Global Search.
Architectural Best Practices & Technical Recommendations
Based on the setup configuration above, keep these key recommendations in mind for a smooth production rollout:
- Data Governance & Data 360 Verification: Before turning on external Data Sources (Step 6), verify that field-level security (FLS) and data masking policies are configured in Data 360. Agentforce Coworker respects underlying data access permissions, so mapping data structures correctly beforehand prevents accidental data leakage.
- Permission Management via Groups: Use Permission Set Groups (Step 11) rather than individual permission set assignments. This scales better for enterprise deployments, simplifies auditing, and keeps user provisioning clean.
- Staged Rollout & User Acceptance Testing (UAT): Step 7 (“Turn on End User Experience”) is marked as optional during initial setup. Keep this disabled in sandbox and pre-production environments until administrative testing is complete via the Ask button next to Global Search (Step 15).
- Asynchronous Activation Patience: Do not interrupt the session during Step 4. The underlying platform provisions several background services and schema links during this progress phase; premature navigation can lead to incomplete data source bindings.