This reference guide provides information for configuring and troubleshooting Agentforce Coworker. It includes:
Agentforce Coworker is an autonomous AI teammate that embeds directly into every Salesforce search bar. Unlike passive search tools, it acts as a proactive partner capable of conversing with all CRM data, reasoning over complex requests, creating multi-step execution plans, and invoking backend tools to complete tasks autonomously.
Q: What data sources can be connected to Agentforce Coworker?
Salesforce first-party sources (CRM and Slack) work out of the box. Over 270 external sources are available via Data 360 connectors — including zero-copy connectors, hyperscalers, data lakes/warehouses, knowledge bases, file stores, collaboration tools, and enterprise apps. Permission-aware sources (Google Drive SharedDrives, SharePoint, Confluence coming soon) can also be linked with inherited source permissions.
Q: What specific data sources are searchable during Beta?
Salesforce CRM (Accounts, Opportunities, Contacts, Cases, Custom Objects, and more)
Slack (messages, channels, and conversations the user has access to)
6 DMOs: Salesforce Product Descriptions, SE Competitive Intel, Sales Content DMO, SE Customer Stories, SE Competitive Intel Source, Sales Content Taxonomy
All Data 360 connectors
Q: Is data from external sources ever stored/saved?
Yes — data from non-zero-copy external sources will be ingested and stored within Data 360. All stored data inherits the full compliance controls and guardrails available in Data 360.
Q: Are there any file size limitations?
Yes. Files over 75MB will not be indexed or brought into search.
Q: How scalable is Agentforce Coworker?
Built on the Agentforce 360 platform, it supports petabyte-scale data volumes and thousands of concurrent queries per second (QPS) per tenant. System limits will be enforced during Beta and initial GA — final limits will be communicated closer to GA. Refer to current Data 360 Limits and Guidelines for per-document and index size details.
Q: Is Agentforce Coworker connected to Calendars? Can it book meetings?
Calendar connectivity (e.g., Google Calendar) is not available out of the box in Beta. The "book a meeting" use case refers to orchestrating a "Book a Meeting" Agentforce action — not a direct calendar integration. Customers who want calendar data searchable must connect Google Calendar separately via a Data 360 connector.
Q: How does Agentforce Coworker handle multilingual enterprise search?
Beta: English only. Other languages may partially work but have not been comprehensively tested.
GA: Will support all languages that Salesforce and Agentforce support.
Before activation, admins must ensure the organization is on an A4X / A1E plan or has Flex Credits allocated for usage-based metering visible in the Salesforce Digital Wallet. The following must be explicitly assigned:
Permission Set License (PSL): "AI Search Setup" (or updated branding equivalents).
Permission Set / Group: "Ask Agentforce Admin" and Access_Ai_Search assigned directly to users.
Official Dev Docs – Turn On Instructions
Navigate to Setup → Agentforce Coworker (or Setup → Set Up Data Space).
Click Get Started and toggle Turn On to provision the environment.
Manage Data Space (For Multi-Data-Space Orgs): Coworker must be bound to exactly one dedicated Data Space chosen at activation.
Manage Sources: Select and index the Data Cloud Model objects (DMOs), Unified DMOs (UDMOs), or federated connectors.
Manage Users: Map access permissions to the intended user groups.
Q: Does Agentforce Coworker work with DC1 (Companion Org setup)?
Yes — Agentforce Coworker can be used on a DC1 setup.
An admin must first set up Agentforce Coworker in the Home Org, including any Data 360 data sources to be used.
After successfully completing that step, Agentforce Coworker can be turned on in the Companion Org(s).
Q: Is Agentforce Coworker available for authenticated Experience Cloud users?
Yes.
Q: Is Agentforce Coworker available in GovCloud?
Not yet. Work is in progress and availability will be communicated once ready.
No, for users with Unmetered User-based AI Entitlements, their usage will show 0% consumption in Digital Wallet and won’t count towards their contracted usage limit.
|
Error Code / Message |
Probable Root Cause |
Resolution Steps |
|
401 Unauthorized |
The underlying Connected App or OrgJWT authentication handshake is misconfigured or detached. |
Verify Connected App definitions in Setup and ensure the OrgJWT certificates match. |
|
400 Invalid ID |
An incorrect or poorly truncated Agent ID format was passed to the session API endpoint. |
Extract the exact 18-character Agent ID string directly from the Salesforce setup URL. |
|
"Not Enabled" / Permission Mismatch |
The specific target user lacks the necessary permission set licenses or permission group assignments. |
Audit user assignments for Access_Ai_Search and confirm the license is active. |
|
500 Server Error / Engine Lookup Failure |
Temporary backend platform failure or an interruption in the LLM Gateway connection. |
Perform a hard browser cache refresh, and if needed, toggle Agentforce Coworker off and on in Setup. |
|
Credential Error / "Invalid Token" |
The session initiation payload failed to pass a valid OAuth/JWT token to the conversational endpoint. |
Verify client credentials and gateway network paths. |
Refer back to the Setup, Licensing & Prerequisites section.
Agentforce Coworker fully honors the existing Salesforce access model. The system recognizes user identity across the enterprise and enforces source permissions through Attribute-Based Access Control (ABAC) — search results and agent actions respect the same access controls as the underlying systems.
Q: Do users need to explicitly grant access to external sources?
Permission-aware sources (Google Drive SharedDrives, SharePoint — Confluence coming soon): No explicit grant needed; policies are automatically created. Note: currently in a closed pilot — must be enabled per org on request.
Non-permission-aware sources: Yes — policies must be manually configured using Data 360 Governance capabilities.
Q: How are external data permissions kept up to date?
Federated sources (queried at runtime): Permissions enforced by the source app in real time — no ingestion into Data 360.
Ingested permission-aware sources: Data 360 ingests from the source, converts to a Data 360 object, and automatically creates and applies policies. Source permissions are inherited.
All other connectors: Permission model is not inherited — governance must be created and applied manually via Data 360 governance capabilities.
Q: What out-of-the-box governance controls are included?
Enterprise-grade controls are built in via Data 360: PII masking, toxicity detection, and result grounding are available out of the box. Customers can also create custom governance policies via Data 360 governance capabilities.
Refer to the following documentation for additional setup, configuration, billing, and user guidance.
Set Up Agentforce Coworker (Beta)
Billing Considerations for Agentforce Coworker (Beta)
005387735

We use three kinds of cookies on our websites: required, functional, and advertising. You can choose whether functional and advertising cookies apply. Click on the different cookie categories to find out more about each category and to change the default settings.
Privacy Statement
Required cookies are necessary for basic website functionality. Some examples include: session cookies needed to transmit the website, authentication cookies, and security cookies.
Functional cookies enhance functions, performance, and services on the website. Some examples include: cookies used to analyze site traffic, cookies used for market research, and cookies used to display advertising that is not directed to a particular individual.
Advertising cookies track activity across websites in order to understand a viewer’s interests, and direct them specific marketing. Some examples include: cookies used for remarketing, or interest-based advertising.