Imagine two situations unfolding at a vehicle dealership.

A customer brings in a vehicle before the monsoon. Its age and odometer reading suggest routine maintenance, but its actual usage pattern, previous complaints and operating terrain indicate that certain parts may wear out sooner.

At the same time, the dealership is preparing its next parts order. Historical averages suggest normal demand, but seasonal consumption patterns indicate that demand for specific parts is likely to rise.

In a conventional DMS, these decisions depend largely on fixed rules, historical averages and individual experience. In an AI-enabled DMS, the system can identify these patterns early and support both service and procurement decisions.

This was the opportunity we addressed while enabling AI capabilities in Excellon DMS ecosystem.

The problem

Two related challenges affected the parts and service lifecycle:

1. Dealers and distributors needed a more reliable way to forecast part requirements and maintain adequate inventory without creating excess stock.

Although these challenges appeared in different processes, they were connected. An inaccurate service recommendation could affect customer experience, while weak demand forecasting could result in the required part being unavailable when the customer visited the workshop.

Root causes

The key limitations were:

• Inventory planning was primarily based on historical consumption and static reorder logic.

• Seasonal and geographical variations were difficult to incorporate consistently.

• Vehicles of the same model and age could have significantly different usage patterns.

• Customer complaints and previously consumed parts were not fully utilized to predict future requirements.

• Fixed maintenance rules could recommend a part too early or overlook an emerging wear pattern.

• Acceptance or rejection of system recommendations was not being used to improve future suggestions.

How Excellon DMS Uses AI for Predictive Parts and Service Decisions

The Excellon team introduced AI-driven decision support at two critical points in the DMS journey.

The first capability focused on AI-based part purchase forecasting. The model analysed historical part consumption across service invoices, counter sales, dealer sales, returns and stock transfers. It also considered seasonal behaviour, regional conditions, terrain, lead time, available stock, in-transit quantities, open purchase orders and back orders.

The resulting forecast was converted into a suggested purchase quantity and displayed within the dealer’s existing ordering workflow, along with a confidence score.

The second capability focused on AI-based part recommendations during job-card preparation. The model evaluated the vehicle model, age, current odometer reading, service history, earlier customer complaints and previously consumed parts.

An exponential moving average was used to understand the vehicle’s recent running pattern. A heavily used vehicle could therefore receive an earlier preventive replacement recommendation, while a lightly used vehicle would not automatically receive the same recommendation merely because it had crossed a predefined age or odometer threshold.

Seasonal and terrain-related wear patterns were also incorporated into the recommendation logic.

High-level solution logic

The implementation followed a simple decision flow:

For purchase forecasting, predicted demand was adjusted against the available and expected inventory position.

For job cards, the AI model generated the top relevant part recommendations, displayed only those meeting the defined confidence threshold and arranged them in descending order of confidence.

The system also captured whether recommended parts were accepted, deleted or rejected. This created a feedback loop to improve the accuracy and relevance of future recommendations.

Importantly, AI supported the user’s decision it did not remove human judgement from the process.

Business Outcome Across the Automotive Ecosystem

The implementation created value across the automotive ecosystem.

For the OEM, it provided better visibility into potential parts demand, supported improved network-level planning and created a foundation for comparing AI recommendations with OEM-defined preventive-maintenance standards.

For dealers and distributors, it enabled more informed ordering, reduced dependence on manual estimation, supported healthier inventory levels and improved the probability that the right part would be available when required.

For vehicle owners, it supported more relevant service recommendations, reduced avoidable replacement suggestions and lowered the risk of repeat visits caused by unavailable parts. The broader outcome was a faster, more transparent and more dependable service experience.

The real value of AI in a DMS is not simply its ability to predict. It is its ability to connect service behaviour, customer needs, inventory availability and procurement decisions turning operational data into timely, practical action.