Contact us
A global automotive logistics company moves hundreds of thousands of finished vehicles across North America each year, and dealers could only find out where a car was by phoning carriers or operations. We built its first real-time tracking portal for dealers, fed by live carrier telemetry, with the Oracle system of record left running underneath.
A buyer in the showroom, and nobody knew where the car was.
Delays came to light only after a dealer had felt them.
As the company reported.
The company moves finished vehicles from port to dealer for several car brands. When a buyer asked when their car would arrive, the dealer had to phone the carrier, email operations or wait for an overnight report. Vehicle, shipment and carrier information lived in separate systems, so every question meant a search across several of them.
Shipping decisions were steered by last week's report rather than live data, so a delay was usually found after it had already caused a problem. Dealer agreements needed strict separation, with each dealer seeing only its own vehicles, and the old platform couldn't enforce that architecturally. Adding a carrier or telematics provider meant custom engineering against systems that weren't built for it.
We left the warehouse and dispatch systems that ran the business in place and built a new interface onto them: a VIN-level portal for dealers, fed by live telemetry, that reads from the systems of record without disturbing them.
3 decisions, in the order we made them. Each one says what we did and why, with what it did drawn beside it.
Keeping Oracle as the system of record let the warehouse and dispatch teams keep working through the whole project, with zero downtime to their processes.
Lambda processes carrier telemetry from IoT-equipped carriers and normalizes it for the application. DynamoDB stores the live position and status the portal reads, at single-digit-millisecond latency. Shipment records stay in the relational systems; the dealer view reads changing position and status from DynamoDB.
SNS and Lambda trigger notifications when delivery status, exceptions or the expected arrival change, with email through SES, so a dealer hears about a delay before the buyer asks. Operations teams see carrier performance and delay patterns from the same live data. Adding a new IoT-equipped carrier is configuration on the same pipeline.
Why: position changes constantly and shipment records rarely do, so keeping them in different stores means neither slows the other down.
The platform runs inside an isolated VPC, Secrets Manager handles integration credentials, and CloudWatch and CloudTrail cover operations and activity review. An automated CI/CD pipeline ships new services without manual coordination.
As the company reported.
More carriers and telematics partners onboarded onto the same pipeline, and continued work on the company's roadmap.
This project came before our current offers. Here's how we'd approach the same work now.
With the data already live, an AI agent could answer dealer and buyer questions directly, on Amazon Connect or in chat, signed as a Guaranteed AI Outcome against a number the business tracks, such as status calls handled without a person, with go or stop at week 4 and payment on the result after it. Under AK AI Factory's Control job, acting for a dealer, it would get that dealer's rights and no more, through Cedar policies checked outside the model on every call, and each tool, a VIN lookup or an ETA, would carry its own permission.
Predicted ETAs would be proved before anyone saw them: a golden set of past shipments with their real arrival times, a sealed set beside it, and a predicted ETA shown only if it beats the current estimate on both.
A free workshop: a few hours with the architects who'd do the work, on your own systems. No cost.
Book a free workshop