Hemut Is Bringing Trucking Out of the Pen-and-Paper Era With AI Agents

The Macro: A $900 Billion Industry Running on Paper and Phone Calls

Trucking is the backbone of the American economy. Over 70% of all freight in the United States moves by truck. The industry does about $940 billion in annual revenue. And a staggering amount of it still runs on phone calls, faxes, handwritten logs, and spreadsheets held together by duct tape and institutional knowledge.

I am not exaggerating. Talk to anyone who operates a small trucking fleet and they will describe a back office that looks like it is stuck in 1997. Dispatchers juggling multiple phone lines. Paper BOLs (bills of lading) getting faxed back and forth. Invoice data being manually keyed into QuickBooks. Driver schedules managed on whiteboards. Rate confirmations buried in email chains that nobody can find when a dispute comes up three months later.

The big carriers have solved some of this. JB Hunt, Werner, Schneider, and the other publicly traded fleets have invested millions in TMS (transportation management systems) from vendors like Oracle, SAP, TMW, and McLeod. But those platforms are designed for companies running thousands of trucks. They are expensive, complex, and require dedicated IT staff to maintain.

The small and mid-size carriers, which is where most of the industry actually lives, are underserved. There are roughly 900,000 trucking companies in the US, and the vast majority of them operate fewer than 20 trucks. These operators do not have IT departments. They do not have six-figure software budgets. They need something that works the way they already work, which means handling phone calls, processing documents, and keeping the books straight without requiring a computer science degree.

Several companies have tried to modernize this space. KeepTruckin (now Motive) went after ELD compliance and fleet tracking. Rose Rocket built a modern TMS for mid-size carriers. Relay Payments digitized carrier payments. But the back office, the actual daily operational chaos of running a trucking company, remains largely untouched by software.

The Micro: A Founder Who Lived the Problem

Hemut is a three-person team out of San Francisco building what they call an AI-powered operating system for trucking companies. The founder is Loki Cheema, and his path to this company is one of the more compelling founder stories I have come across this year.

Cheema became homeless at 16. He worked 100-hour weeks at a motel, where he met truck drivers who needed help with their paperwork. That turned into a dispatching side business, which turned into a full dispatching company. He spent six years in logistics, doing the exact work that Hemut is now trying to automate. When voice AI technology matured enough to handle real phone conversations, Cheema saw the opportunity to fix the bottlenecks he had personally experienced for half a decade.

The product has four main components. First, AI agents that handle inbound and outbound calls. In trucking, the phone is still the primary communication channel. Brokers call to offer loads. Shippers call to check on deliveries. Drivers call about everything. A small fleet might field dozens of calls a day, and missing one can mean losing a load worth thousands of dollars. Hemut’s voice agents handle these calls, extract the relevant information, and route it into the system.

Second, document processing. Trucking generates an absurd amount of paperwork: rate confirmations, BOLs, proof of delivery, insurance certificates, invoices. Hemut lets you upload these documents and uses AI to extract the data, eliminating manual entry. If you have ever watched a dispatcher squint at a faxed rate confirmation and type numbers into a spreadsheet one field at a time, you understand why this matters.

Third, load sourcing. Finding profitable loads is the central challenge of running a trucking company. Hemut helps operators find and evaluate available loads, which today involves checking multiple load boards, calling brokers, and comparing rates manually.

Fourth, automated accounting. Getting paid in trucking is notoriously slow and messy. Hemut handles invoicing and financial tracking, connecting the operational data (what was hauled, when, for how much) to the accounting side without manual reconciliation.

Cheema came through Y Combinator’s Spring 2025 batch. The company is early, but the product scope is right for the market. Small fleet operators do not want five different software tools. They want one thing that handles the whole back office.

The Verdict

I think Hemut is attacking exactly the right problem at exactly the right time. Voice AI has reached the point where it can handle real business calls reliably. Document understanding models are good enough to parse the messy, inconsistent paperwork that trucking generates. And the market is enormous, fragmented, and chronically underserved by existing software.

Cheema’s background is the strongest signal here. This is not a Silicon Valley founder who read a McKinsey report about logistics and decided to build a startup. This is someone who spent six years dispatching trucks, answering those phone calls, filing that paperwork, and chasing those payments. He knows what the daily pain feels like because he lived it.

The competitive risk is not from the enterprise TMS vendors. Oracle is not coming downmarket to serve a 10-truck fleet. The risk is from other AI-native startups targeting trucking ops, and from Motive potentially expanding from fleet tracking into back-office automation. But Hemut’s integrated approach, handling voice, documents, loads, and accounting in one system, is harder to replicate than any single feature.

What I want to see in 90 days: how many fleet operators are using the voice agents for real calls with real brokers, and whether those operators are also using the document processing and accounting features or just cherry-picking one module. If the whole system works together, the switching costs become very real very fast. That is how you win in vertical SaaS.