By Dave DeFusco
On any given day in the United States, thousands of truck drivers spend hours scanning online load boards, searching for freight they can haul. The process is often slow and frustrating, requiring drivers or dispatchers to sift through messy data, call brokers and negotiate prices. For Tharun Prabhakar, a graduate of the Katz School鈥檚 M.S. in Artificial Intelligence, that inefficiency looked like an opportunity. Today, he is helping build technology that could transform how freight moves across the country.
At the Katz School, Prabhakar took courses taught by computer scientist Jiang Zhou. After graduating, he became a founding member of Kayaan AI, a startup developing voice-driven artificial intelligence tools for truck drivers. The goal of the company is simple to explain but technically complex to achieve: create an AI assistant that allows a truck driver to book a profitable load with a single voice command.
鈥淚nstead of drivers spending hours searching and negotiating, our system does the work in the background,鈥 said Prabhakar. 鈥淵ou just talk to it, and it understands what you want, finds the best load and even negotiates for you.鈥
At Kayaan, Prabhakar works as what is known as a Forward-Deployed AI Engineer, a role that blends technical development with hands-on work with customers.
鈥淢y work doesn鈥檛 involve just sitting behind a screen and coding all day,鈥 he said. 鈥淚 talk directly with clients, understand their requirements and filter out the noise. Then I quickly build a proof of concept so they can see how the system will work before we develop the full product.鈥
The platform relies on several artificial intelligence systems working together. One central AI agent鈥攏icknamed 鈥淜ay鈥濃攃ommunicates with drivers through voice. Other AI agents handle specialized tasks behind the scenes.
鈥淲e have two or three different AI systems running behind the app,鈥 said Prabhakar. 鈥淭he main agent talks to the driver and understands their needs. When the driver finds a load they like, another agent is triggered to call the broker in real time, confirm the details and negotiate the price.鈥
The negotiation process can resemble a real conversation between two people. 鈥淚t goes back and forth,鈥 said Prabhakar. 鈥淚t pushes the boundary as much as possible to get a better rate for the driver. At the same time, it keeps the driver informed and asks for approval before the load is booked.鈥
Building such a system has required solving practical challenges that don鈥檛 always appear in academic research. One of the toughest was teaching the AI to function in noisy environments.
鈥淭he biggest task was making it work in noisy conditions,鈥 said Prabhakar. 鈥淎 truck cab can have multiple voices, road noise and radio. We had to build tools so the system can recognize who the driver is and ignore background voices.鈥
The team also had to make sense of the messy data that dominates the freight industry. 鈥淎I alone is not enough,鈥 he said. 鈥淢ost of the data we get is very unstructured. So, we first build algorithms that extract meaning from the data and convert it into a format the AI can understand.鈥
The business vision behind Kayaan is supported by fellow Katz student Aquib Hussain, a student in the M.S. in Data Analytics and Visualization and contributes to the company鈥檚 strategy.
鈥淥ur vision is to build a voice-first AI co-pilot for truck drivers,鈥 said Hussain. 鈥淩ight now drivers depend heavily on dispatchers and often deal with unfair prices. What we鈥檙e building helps drivers book profitable loads themselves with a single voice command. The core team is supported across analytics, design thinking, customer insights, product feedback and go-to-market strategy as we bring this platform to life.鈥
The scale of the opportunity is enormous. The trucking industry in the United States alone moves more than a trillion dollars鈥 worth of goods each year. Even small improvements in efficiency can have huge economic impact.
Prabhakar鈥檚 journey from graduate student to startup founder reflects the growing connection between academic research and real-world technology deployment.
鈥淚t鈥檚 more than the curriculum,鈥 he said of his time at the Katz School. 鈥淭he projects and the professors guided me in understanding how real systems work. They taught me how to break complex problems into smaller pieces and build solutions step by step.鈥
Professor Zhou said watching students like Prabhakar translate classroom ideas into working technology is deeply rewarding.
鈥淭he initiative that Tharun and his teammates have shown is very impressive,鈥 he said. 鈥淭hey are taking advanced AI concepts and applying them to a real industry problem. That kind of innovation is exactly what we hope our students will pursue.鈥
