TrueMile puts AI dispatch to work inside its own trucking fleet

TrueMile is using its own trucks to test a different operating model for freight. The company combines mathematical optimization with AI agents that find loads, contact brokers and coordinate dispatch, while a person still approves every load and signs every rate confirmation.

Founder Harpreet Dhaliwal began running his family's trucking business at 19. According to the company's launch, he restructured debt, upgraded the fleet and returned it to profitability. The tools developed for that operation became the basis for TrueMile, now part of Y Combinator's Fall 2026 batch.

The problem it addresses is less obvious than choosing the load with the highest advertised rate. A well-paid trip can leave a truck in the wrong place for the next commitment, or require a costly empty return. Dispatchers have to balance those tradeoffs while dealing with delays, cancellations, available driving hours and changing freight demand.

TrueMile says its planning system evaluates contracts, load assignments and return freight together. It weighs forecasts, revenue, costs and driver commitments, then uses agents to find suitable loads and negotiate with brokers. Other agents handle dispatch and customer updates, with assignments revised when conditions change.

Watch the original TrueMile demonstration.

The company reports that, over four months, AI-selected loads earned 17% more revenue per mile than the market average for dry vans and 8% more for refrigerated trailers. It also says its trucks averaged 10% more miles per month than the industry average. These are company-reported comparisons. The announcement doesn't identify the benchmark dataset or provide enough detail to separate the effects of routes, equipment and freight mix.

Revenue per mile also isn't the same as profitability. The operational case depends on costs and utilization as well as rates, which is why the whole-fleet planning approach is more consequential than the headline percentages alone. TrueMile's stated goal is to keep trucks productive without increasing office headcount at the same pace as the fleet.

The company is starting with human drivers. Its public YC profile describes autonomous trucks as a future possibility when they become available on its routes, rather than a current capability. For the present product, the human approval requirement remains part of the operating process even as agents perform the coordination work.

Dhaliwal holds a master's degree in data science. Founding engineer Daniyar Abykhanov previously built voice agents at a large bank in Kazakhstan, according to the launch. The company's public materials identify him as a founding engineer, not a co-founder.

Running the fleet gives TrueMile a direct connection between a software recommendation and its outcome on the road. It also means the company is responsible for the service it sells: trucks, drivers and delivery commitments remain central to the business. Customers are buying freight capacity backed by this operating system, rather than simply purchasing a dispatch tool for their own fleet.

TrueMile is seeking two dedicated-lane partners for recurring truckload freight in Texas and the Sunbelt this quarter. That is the immediate expansion plan described in the launch: repeatable routes and commitments that its planning system can optimize together, with a person retaining the final say on each load.