Hemut: Building the Internet of Freight on Real-Time Data

Everything Americans buy spends part of its life on a truck. Groceries, lumber, medicine, and packages delivered to our door all depend on trucks at some point in their journey, meaning nearly every dollar a consumer spends ultimately helps pay to move something by truck. When trucking runs inefficiently, everyone pays for it. And trucking remains remarkably inefficient.

The companies hauling America’s freight are mostly small businesses operating on thin margins, with back offices working nights and weekends simply to keep up. For decades, the industry has lacked software capable of taking on much of that work. The consequences ripple outward as family businesses struggle to survive, drivers wait to get paid, and consumers ultimately bear the cost through higher prices for nearly everything they buy.

An AI-Native Operating System for Trucking 

Hemut, a Y Combinator Spring 2025 company, was founded on a different premise: a trucking company does not have to operate the way it has for the past fifty years. When AI can answer the phones, keep the books, handle paperwork, and monitor every truck around the clock, a fleet can operate at a structurally lower cost without relying on a back office working nights and weekends to hold everything together.

That is Hemut’s product today: an AI-native operating system for carriers and brokers, combining a full ERP and TMS into a platform where much of the work runs itself. The longer-term vision is bigger. Hemut calls this vision the Internet of Freight, an industry in which fleets operate on a connected network that allows efficiency gains to compound and benefit everyone the industry touches, including carriers whose costs fall, drivers who get paid faster, and consumers who pay less for the goods trucks deliver.

But none of that works if the underlying picture of the business is not current. An AI system acting on stale information can make matters worse rather than better. That is why Hemut joined Cohort-3 of Confluent’s Data Streaming AI Accelerator in February 2026: to build a real-time data foundation for its platform with Confluent Cloud. That foundation has been in production since March 2026 and today supports every fleet Hemut has brought onto the platform.

A Front Desk in the Pacific Northwest 

Hemut founder and CEO Loki Cheema didn’t come to trucking through a supply chain degree or a family business. He came to it through a motel front desk. As a teenager working at an Everett, Washington motel that catered to long-haul truckers, Cheema began helping drivers with their paperwork: bills of lading, logbooks, settlement statements, the mountain of forms every trucker has to keep straight to get paid and stay compliant. It earned him a little extra money, and by the time he finished high school, it had given him a firsthand education in how fragmented and manual American freight really is.

“I watched drivers at that front desk lose money to paperwork that never talked to anything else,” Cheema said. “The truck was never the problem. The information was the problem. Hemut exists so a carrier runs their whole operation off one live picture instead of five systems that find out what happened a day late.”

After graduating from UCLA with degrees in computer science and linguistics, Cheema set out to build what the truckers at that front desk never had: a single operating system designed around how freight actually moves, and behind it, a freight industry that finally works like a network.

From Software You Watch to Software That Works

The last generation of trucking software sold visibility: dashboards that showed operators what was happening across their business. Hemut’s bet is that the next generation will sell something more valuable: labor. Instead of software that simply shows you the operation, Hemut is building software that actually does the work.

That vision is already in production. At fleets running on Hemut, AI voice agents answer the phone 24/7 with the truck’s live location and the fault code that just fired already in front of them. Automations carry back-office workflows from beginning to end, with settlements preparing themselves, invoices going out as soon as a delivery is confirmed so cash comes in sooner and drivers get paid faster, and the books reconciling as money moves.

The result is that the back office gets its nights and weekends back, while dispatchers can spend more of their time on the two relationships that actually run a trucking company, their drivers and their customers. Hemut also connects directly with brokers’ and shippers’ systems, allowing load paperwork that once had to be manually rekeyed to flow automatically. The dispatcher, the truck, and the ledger all operate from the same picture of the business at the same time. Each fleet’s data remains its own, with everything a carrier streams into the platform visible only to that carrier and no one else.

Every one of these capabilities shares the same fundamental dependency: the AI needs to know what is happening as the operation changes. The nightly batch cycles on which trucking software has traditionally relied are precisely what keep software passive. Eliminating that lag was therefore one of the first things Hemut set out to do.

“Margin in trucking is decided in the moment,” Cheema said. “Every hour between the data and the decision costs a carrier real money. That gap is what we are taking out.”

Why Hemut Built on Confluent

For Hemut’s engineering team, the infrastructure calculus was straightforward. A lean team building an operating system for an entire industry should not spend its time managing streaming infrastructure that is not the product.

Without a dedicated infrastructure team, Hemut used the Data Streaming AI Accelerator program to build its real-time data foundation on Confluent Cloud and put it into production within weeks. Five months later, the results are measured directly from the platform, not estimated. The system has carried more than 286 million events without losing a single one.

Every truck and trailer on the platform continuously streams operational data: live location, stops and idle time, fuel economy, odometer readings, tire pressure, and a steady feed of engine diagnostics. A position update moves through the platform to every service that needs it in roughly a tenth of a second.

“Every truck is basically a sensor network on wheels,” said Hemut founding engineer Sumedh Kane. “The moment we can treat that as a real-time stream instead of a batch export, we can start doing things carriers have never had access to before: predictive maintenance that’s actually predictive, and routing decisions that account for what the truck is telling us right now.”

Just as important as speed is what the platform makes routine. Hemut has onboarded new telematics providers and evolved its data formats multiple times with zero downtime and without breaking downstream consumers, with Confluent’s Schema Registry enforcing the data contracts. And because the platform retains a rolling month of history, when Hemut improves a predictive-maintenance model, the team can rerun it against a full month of real-world fleet data without having to recollect a single data point.

Confluent Cloud allows Hemut to run all of this while keeping its engineers focused on the product. As the network expands, the team can add Flink for stream processing and build on that same real-time data foundation for AI without re-platforming the underlying architecture.

What began with telematics data is now becoming the real-time architectural pattern for the entire operating system, connecting dispatch, accounting, sales, and brokerage through the same continuously updated data foundation.

The Payoff: Hours Back, Margin Back

Hemut’s platform is already live with real fleets, working alongside some of the industry’s largest incumbents while also serving the mom-and-pop operators that have historically lacked access to sophisticated technology and automation.

At one of Hemut’s largest deployments, the team spent a summer embedded inside the carrier’s operation, working alongside every department to understand exactly where time and money were being lost. Across eight core workflows, they identified roughly 37,000 hours of manual work per year – equivalent to approximately $1.3 million in labor costs. Hemut brought the carrier’s full technology stack live in just two days, running it alongside the existing software so that every workflow could be tested and verified before the transition.

At Hemut’s anchor fleet, which now runs its day-to-day operations on the platform, quoting, scheduling, tracking, and asset agents, along with automations spanning the rest of the business, handle much of the routine work around the clock. The people who once spent their days buried in administrative tasks can instead focus their time on the two relationships that matter most – customers and drivers.

The same principle extends to the trucks themselves. Hemut’s telematics models are designed to extend vehicle service life and identify mechanical problems before they become costly roadside breakdowns, acting on what each truck reports as operating conditions change.

“Carriers have been sold hardware-centric telematics for years,” Kane said. “What we’re building is different. It’s an operating system where the telematics stream, the ERP, and the AI layer all see the same real-time truth at the same time. Confluent helps Hemut turn continuously changing operational data into a real-time, trustworthy foundation for the applications and AI services running the business.”

The destination is the vision Cheema first began forming at that motel front desk, now extended across an entire industry. Hemut’s Internet of Freight allows carriers of any size to operate with the leverage and sophistication of a company ten times larger, powered by a connected network that does more of the watching, coordinating, and work. At its foundation, this new freight network is real-time and event-driven.

Fleet operators interested in seeing what Hemut can do for their own operations can visit Hemut’s website. Early-stage companies interested in building similarly ambitious AI applications on real-time data can learn more about Confluent’s Data Streaming AI Accelerator.

About Hemut

Hemut is building the Internet of Freight: an AI-native operating system for trucking, with voice agents, telematics intelligence, and automated freight bidding on one connected platform. Founded by Loki Cheema and backed by Y Combinator (Spring 2025), Hemut serves small and mid-sized carriers and brokers across the United States. 

Learn more at Hemut.com.