Considerations for Autonomous Scheduling for Field Service

Customer Considerations and Limitations

• Data privacy: The agents can’t modify customer and contact data.
• Past appointments: Customers can’t reschedule or cancel appointments for dates that are in the past.
• Dispatched appointments: Customers can’t cancel or reschedule appointments that are in the Dispatched status.
• Number of requests: Customers can make one request at a time. For example, you can’t ask to cancel all appointments for today in a single request, but you can ask to cancel one at a time within the same conversation.
• No response: After sending the initial outreach message, if the customer doesn’t respond, we wait before sending a reminder:
  • When rebooking a scheduled appointment, the reminder is sent 2 hours after the initial message.
  • When booking an appointment for Asset Maintenance, the reminder is sent after 80% of the time has passed between the due date and the initial message.

  In both cases, after the customer gets the reminder, if they still don’t respond, we wait for an additional 2 hours before escalating the conversation.

• Reminders: If a customer receives an outreach message and asks for a reminder, Agentforce can’t provide this service.

Agent Considerations and Limitations

Before you begin, review the considerations for Agentforce Service Agent.

• Additional agent subagents:
  • The Reservation Management subagent isn’t included by default in the Scheduling agent but is included in the Agentforce Service Agent. It’s possible that this subagent conflicts with the Appointment Management for Field Service subagent. If you choose to include both subagents in your agent, make sure to test the behavior and if issues occur, remove the Reservation Management subagent from the agent.
  • Don’t add the Field Service Dispatcher Actions subagent to the Agentforce Service Agent or Scheduling agent. The subagent is only supported by the Employee agent.
  • The agent can either work with the Appointment Management for Field Service subagent or with the Appointment Management for Scheduler subagent, but not both.
• Agent user: Service appointments created by Agentforce are marked as Created By and Last Modified By the agent user selected when creating the agent.
• Customer Outreach:
  • 100 outreach messages can be sent to customers per customer outreach request.
  • Conversations that start with the outreach message trigger the Appointment Management for Field Service subagent. The Appointment Management for Scheduler subagent isn’t supported.
  • If the rescheduling request is initiated from the Agentforce panel, it must include the date and only one of these parameters: resource name, contact name, work type, or service territory. Multiple parameters can cause incorrect rescheduling. However, multiple instances of the same parameter can be included. For example, “Reschedule all of today’s appointments for North Los Angeles and East Los Angeles". See Sample Utterances for additional phrases you can use.
• Email: Email isn't supported for Autonomous Scheduling agent conversations.
• Messaging channels:
  • Inbound conversations: Field Service Scheduling agents for inbound conversations can connect to any messaging channels that Agentforce Service agents support.
  • Outreach conversations: Field Service Scheduling agents for outreach conversations can connect to any messaging channel that Agentforce Service agents support except for the standard and enhanced Facebook Messenger channel. The default messaging channel is SMS. To use a different channel, see the Messaging Channels for Customer Outreach section in Configure Autonomous Scheduling.
• Time: If the appointment doesn't contain an arrival window, the scheduled start time is used instead, and the time displayed is rounded up or down to the closest half hour.

Field Service Setup Considerations and Limitations

Review these considerations and make changes as needed.

• Address validation: For the address to be considered valid, the contact must have a Mailing City and a Mailing Street. If one of them isn’t valid, the conversation is escalated to a representative.
• Created service appointment: When getting slots for appointments, the getSlots API requires a service appointment with a parent work order. The Appointment Management for Field Service subagent creates the service appointment and the parent work order. If the service appointment that was created isn’t used in the end, it still remains in the system and isn’t automatically deleted. See AppointmentBookingService Class.
• Email: If different contacts have the same email address, the agent refers to the first contact it finds.
• End of Time Horizon: The end of the time horizon for appointments offered to customers is based on the work type’s Due Date Offset property. If this property is empty, the due date of the created service appointment is set to 7 days. When searching for time slots for this appointment, the system also considers the value defined in this setting: Maximum days to get candidates or to book an appointment. If the value is earlier than the appointment’s due date, this is used instead. The maximum value is 31. To find this setting, go to the Field Service Settings app and open the General Logic tab.

  Examples:

  • If the appointment’s due date is in 60 days and the maximum days to get candidates or to book an appointment is set to 15, the customer is offered appointments within 15 days.
  • If the appointment’s due date is in 20 days and the maximum days to get candidates or to book an appointment is set to 30, the customer is offered appointments within 20 days.
• Optimization: Even if you aren’t currently using optimization features in your org, you can still use the agent for Autonomous Scheduling. If you haven’t configured the optimization engine settings, the default values for Scheduling Policy and Operating Hours are used. See Configure Autonomous Scheduling for Field Service in the Legacy Agentforce Builder. If you have configured the optimization engine, your settings are used instead.
• Service territories: The agent requires a service territory classification mechanism, such as map polygons, to assign service territories to service appointments. Each service territory must also have active service territory members.
• Time zone:
  • The AI agent assumes that the end-customer’s time zone matches the service territory’s time zone. Appointment times are shown in the service territory’s time zone. Make sure that you assign service territories to service appointments. If you don’t, the time zone of the service appointment is taken from the operating hours.
  • When an agent requests appointment slots using a relative date expression, such as 'tomorrow’, the system resolves the time based on the agent's current time, not the service appointment's current time. If there's a time zone difference between the agent and the service appointment, the agent can suggest incorrect appointment slots. If you encounter issues with time zone resolution, make sure that your agent has the Get Current Time for Field Service agent action in the Appointment Management for Field Service subagent.
• Work Rules: The default scheduling policy is Customer First, which includes the required work rules. These work rules help the agent get the time slots for an appointment. If you’ve modified the Customer First scheduling policy or if you prefer to use a different scheduling policy, make sure that it includes the Scheduled Start and Scheduled End work rules. See Complete Prerequisites for Autonomous Scheduling in the Legacy Agentforce Builder.
• Work types:
  • The AI agent uses the name and the descriptions of your work types to find the most accurate work type for the service appointment. In the work type records, make sure that the names and descriptions are as detailed and accurate as possible.
  • All of the org’s work types are considered for scheduling. To speed up processing, you can modify the list of work types to only show the relevant ones. See Configure Autonomous Scheduling for Field Service in the Legacy Agentforce Builder.