Context Service acts as a layer between the application and the rules or procedures, providing the requested data during both the configuration (design-time) and execution (run-time) stages. Context definitions are the key component of the Context Service. By utilizing this layer, ARM agents do not need to perform expensive, repeated database queries; instead, they consume a unified JSON payload of contextual data.
Context definitions eliminate silos by enabling applications like Transaction Management, Product Discovery, and Billing to communicate consistently.
parentReference maps to the custom object's immediate
parent field (e.g., quote_line_item_c).
When an AI agent initiates a quoting or ordering sequence, it utilizes the
Place Sales Transaction (POST) API to create a sales transaction with
integrated pricing and configuration. The API manages the transaction
state via the SalesTransactionContext.
| Property Name | Type | Technical Function |
|---|---|---|
contextId |
String | ID of the context that's created for a session of the sales transaction. If the property isn't specified, the Place Sales Transaction API generates it autonomously. |
isBuiltInTransaction |
Boolean |
Indicates whether a new context ID is created for the sales
transaction (true) or not (false).
|
Salesforce provides standard predefined context definitions that work out-of-the-box, though you can also extend or clone them to meet your unique use cases. These ensure data flows seamlessly across your entire revenue lifecycle.
For the Solutions and Technical Architects, configuring the Context Service is what gives the AI agent its "situational awareness." Without Context Mapping, the correct data sources are not connected to the correct nodes and attributes. Once the context definition and mapping are activated, applications (and by extension, Agentforce AI) can efficiently utilize the mapped data for optimized operations.