Meet Akshat Kannan and William Zhang, college dropouts who launched AI startup Zeroset, here's how it emerged with $5.2 million in funding

Two college dropouts, Akshat Kannan and William Zhang, launched Zeroset to help AI agents understand company workflows in 2026. The startup emerged from stealth in October with $5.2 million in pre-seed funding to develop technology for AI agents working across different business systems.

According to Business Insider report, the company's first product, Nebula, collects information form workplace software and tracks how tasks, decision and activities change over time. Zeroset aims to help AI agents understand the steps involved in a task, rather than simply search through company documents for answers.

The funding round was co-led by Gradient Ventures and 2048 Ventures, with participation from Leblon Capita. The company plans to use the money to hire researchers and engineers, support early deployments with businesses and train its AI models.

Who are Akshat Kannan and William Zhang?

Kannan, 18, is the co-founder and chief executive of Zeroset. He left Stanford University in 2026 to work on the startup. Zhang, the co-founder, left the University of Texas in 2025.

Notably, Nebula connects to software use by businesses and tracks changes in information and activity across these systems to build a picture of how work progresses.

By tracking the sequence of events and decisions, the system is designed to help AI agents understand how similar tasks were handled previously.

Why Zeroset want to improve AI agents?

Companies are exploring AI agents to automate work that involves several steps. However, each organisation has its own processes, approval systems and internal practices.

Zeroset wants to build models that understand these differences. Its goal is to help AI agents use information about a company's operations to complete tasks with less human involvement.

How will Zeroset use its $5.2 million funding?

The company plans to use the funding to expand its team, work with early business customers and cover the cost of training its models.

Training is one of its major expenses. Kannan told Business Insider that one training run cost about $100,000 over two weeks.

Zeroset had five employees when it announced the funding. Its Nebula product was still in a closed research preview, with access expected to expand gradually rather than through a single public launch.

The startup also plans to combine licence fees with usage-based charges linked to the data processed and activity generated by AI agents. It is entering a field that includes companies such as Mem0 and Zep, which are also developing technology to help AI systems retain and retrieve information.