Peak-season repair dispatch

The scenario and numbers in this demonstration project are simulated data.

Industry: Appliance brand service center
What was built: Agents, skills, semantic models and write tools, built on demonstration data
Where the knowledge came from: The VIP customer's repair email and service contract, plus demonstration data from the CRM, FSM, WMS and ERP systems

A midsummer repair request

On a midsummer morning, a VIP chain store emails a repair request. The store's air conditioning has stopped completely, and the contract promises arrival and repair within 4 hours.
The data for dispatch sits in CRM (customer tier), FSM (technicians and work orders), WMS (parts stock) and ERP (open purchase orders). Other air conditioning repairs wait for dispatch the same morning.

The service contract is in the Drive. The agent reads it and confirms the 4-hour SLA term and the penalty.
The agent uses semantic models to check technician load in FSM, stock in the central warehouse and in technicians' vans in WMS, and open purchase orders in ERP.

The dispatch manager asks in Sindri: "Which air conditioning repairs in the Taipei area need dispatch this morning? Please arrange them."
The agent reads the repair email, creates an urgent work order in FSM and marks the SLA deadline.
The agent uses the SLA countdown skill to work out the time left. The agent uses the dispatch optimization skill to find a technician by certification, area and the day's load.
The technician certified for inverter compressors has one slot left that day. The agent confirms that he can finish before the SLA deadline and reorders the other repairs.
The agent uses the parts availability skill to check the compressor module. The central warehouse has the last unit, and the agent assigns it to the VIP work order.
Another technician's van carries the same part. The agent proposes a transfer back to the central warehouse for another, non-VIP work order.
The open purchase order arrives too late for peak-season demand. The agent works out an emergency purchase order.
Before the technician assignment, the parts transfer, the last unit's assignment and the emergency purchase order are written back to FSM, WMS and ERP, the agent stops for the dispatch manager's approval.

At service centers like this, new repair requests will come in every day. Maintenance contracts will keep expiring and being renewed.
The agent will collect feedback on its answers from users. The feedback will go to Odin.
Wrong answers will become tests. A change will be kept only if scores improve.

Get started

AI feasibility review of a live workflow