Used Truck Pricing Agent

Problem

Used-truck pricing depends on many variables — year, mileage, body type, condition, market dynamics. The business relied heavily on veteran buyers' and sales reps' market sense, which slowed acquisition decisions, made onboarding hard, and left price revisions on slow-moving stock to be addressed too late.

Approach

• Structured both the company's transaction history and public market data into a referenceable corpus
• Built an agent that, given a vehicle description, proposes a market band, recommended acquisition price, resale range, and revision suggestions for slow-moving inventory — each with reasoning
• Set up an operating loop where buyer overrides flow back into the dataset, so edge cases improve the model over time

Outcome

• Faster initial pricing decisions, especially for time-sensitive auctions
• Captured veteran buyers' tacit knowledge as explainable rationale, sharable with junior staff
• Reduced case-by-case variance in pricing outcomes and shortened inventory holding periods