Zeroset, an enterprise AI startup founded by an 18-year-old CEO and a 21-year-old co-founder, has emerged from stealth with $5.2 million in pre-seed funding to help AI agents understand how companies operate, make decisions and manage workflows.
The San Francisco-based startup was founded in January 2026. The funding round was co-led by Gradient and 2048 Ventures, with participation from Leblon Capital. The company’s valuation was not disclosed.
Zeroset plans to use the capital to recruit engineers and researchers, train its AI models, develop its product and support early enterprise deployments.
The startup is addressing a challenge in enterprise AI: agents may be able to retrieve information from company documents, but they often lack a complete picture of how work progresses, why decisions are made and how teams handle similar situations.
Zeroset aims to fill that gap with its platform, Nebula.
What Does Zeroset Build?
Zeroset is developing Nebula, an enterprise intelligence platform designed to give AI agents an updated understanding of how a business operates.
The platform connects with workplace tools such as Microsoft 365, Outlook, Teams, SharePoint and GitHub. It tracks changes in information and activity across these systems, building a record of how workflows develop over time.
For example, a company document might contain the final decision on a project. However, the reasoning behind that decision could be spread across emails, team messages, code changes and internal discussions.
Nebula aims to connect these pieces of information so AI agents can understand not only what a document says, but also how a decision was reached and what happened next.
Zeroset represents these activities as structured, searchable workflow records, which agents can use as context when handling tasks.
The company believes this approach could help agents complete complex business processes more reliably by giving them access to organizational history rather than relying only on individual documents or isolated information.
Who Founded Zeroset?
Zeroset was founded by Akshat Kannan, the company’s CEO, and William Zhang.
Kannan is 18 and left Stanford University in 2026, according to the information provided. Zhang is 21 and left the University of Texas the previous year.
The founders established Zeroset in January 2026 to address what they see as a major limitation in enterprise AI: existing models do not automatically understand the internal processes and decision-making patterns of the organizations where they are deployed.
The company currently has five employees. It plans to use the new funding to expand its technical team, particularly in research and systems engineering.
Model training is another priority. According to the company information provided, training costs reached approximately $100,000 over a two-week period, highlighting the computing resources required to develop its technology.
How Will Zeroset Use the $5.2 Million?
The pre-seed funding will support four main areas of development.
First, Zeroset plans to hire researchers and systems engineers to improve the platform’s technical capabilities.
Second, it will invest in model training and computing resources needed to build and refine its enterprise intelligence technology.
Third, the company will continue developing Nebula, including its ability to capture workflow changes and make organizational context available to AI agents.
Finally, Zeroset plans to support early enterprise deployments, allowing it to test the product in real business environments and evaluate how effectively its technology helps agents perform complex tasks.
The product remains in a closed research preview, with access expanding gradually. The company has not disclosed customer names, revenue figures or other detailed commercial metrics.
Zeroset’s Business Model and Target Markets
Zeroset plans to combine licensing fees with usage-based pricing. Costs are expected to scale with the amount of data processed and the level of agent activity.
The company is initially targeting manufacturing, supply-chain operations and financial research. These areas often involve multiple systems, detailed operational histories and decisions that depend on information spread across teams and software platforms.
If Nebula can reliably connect that information, it could help AI agents manage workflows that require more than retrieving a single answer or following a simple instruction.
However, Zeroset will need to demonstrate that businesses are willing to pay for a separate intelligence layer and that the product delivers measurable improvements in agent performance.
Why Zeroset’s Funding Matters for Enterprise AI
Zeroset’s funding comes as companies explore ways to move AI agents beyond answering questions and assisting employees toward completing multi-step tasks.
One challenge is that enterprise information is often distributed across different applications. Even when an AI model can access relevant documents, it may not understand the sequence of events, decisions and actions that produced the current situation.
Zeroset is betting that giving agents a structured record of these changes can help them make better-informed decisions and carry out workflows with fewer errors.
The startup also faces competition from established enterprise software providers. Companies that already control workplace applications and business data could develop similar capabilities within their own AI products.
Security and access permissions will also be important. Enterprises will need confidence that any system tracking workplace activity respects internal permissions and protects sensitive information.
Zeroset must therefore demonstrate that its cross-platform approach provides enough value to justify the additional software, cost and integration required.
What Comes Next for Zeroset?
Zeroset’s immediate priorities are expanding its engineering team, improving Nebula and supporting its first enterprise deployments.
The next stage will be proving whether its workflow records help AI agents complete complex tasks more accurately and consistently in real business settings.
With $5.2 million in pre-seed funding and a five-person team, the startup has capital to advance its technical approach. However, its customer adoption, revenue potential and ability to compete with features built directly into existing enterprise platforms remain unproven.
The results of its early deployments will help determine whether Nebula can establish itself as a valuable layer of infrastructure for enterprise AI agents.
About Zeroset
Zeroset is a San Francisco-based deep tech startup founded in January 2026 by Akshat Kannan and William Zhang. The company develops Nebula, an enterprise intelligence platform that captures workflow activity and changes across business systems to help AI agents understand organizational processes and decisions.
The startup raised $5.2 million in pre-seed funding from a round co-led by Gradient and 2048 Ventures, with participation from Leblon Capital. It plans to invest the proceeds in engineering recruitment, model training, product development and early enterprise deployments.
