San Francisco-based Beltic, a startup building infrastructure to help businesses verify and control AI agents, has emerged from stealth with $7.3 million in seed funding.
The round was led by Norwest, with participation from Restive Ventures, Oxford Seed Fund and Collide Capital. Angel investors from companies including Coinbase, Legora and Ramp also participated.
The latest financing brings Beltic’s total funding to approximately $8.8 million since the company was founded in 2025.
As AI agents increasingly browse websites, make purchases and perform tasks on behalf of customers, businesses face a new challenge: determining who is behind an agent and what that agent is allowed to do.
Beltic is developing technology around what it calls “Know Your Agent” (KYA), a concept inspired by the “Know Your Customer” processes used by financial institutions.
Its platform is designed to identify AI-agent traffic, verify the person or business behind an agent and check the agent’s proposed actions against rules established by the business.
Companies can configure the system to allow, block or require additional confirmation for specific actions.
Beltic’s founders say many traditional fraud prevention tools were designed around the assumption that a human is directly operating a device.
These systems can rely on signals such as device fingerprints, mouse movements, typing patterns and selfies.
AI agents do not necessarily provide the same signals, creating challenges for businesses trying to distinguish legitimate automated activity from potentially risky behavior.
Beltic’s platform is designed to address this shift by focusing not only on an agent’s identity but also on its behavior and authorization.
For example, a customer may authorize an AI agent to search for flights but not give it permission to purchase a ticket. Beltic can check the proposed action against the business’s rules before allowing the transaction to proceed.
Beltic’s technology is designed to monitor individual actions performed by AI agents.
If an agent behaves unexpectedly, takes a risky action or appears to move beyond the user’s original intent, the platform can flag the activity and require additional confirmation.
The company says its dashboard can distinguish AI-agent visits from other website traffic, including agents accessing a website directly or through connections to other software.
This allows businesses to accept certain agent activity while applying additional checks before sensitive actions such as purchases.
Beltic was founded in early 2025 by Bhatnagar and Mike Allan, who previously worked on different aspects of payments, identity, fraud prevention and compliance.
Bhatnagar spent nearly six years at Coinbase, where she worked on customer onboarding, fraud prevention and compliance products.
Allan previously founded Atar, a banking-as-a-service company that built payment infrastructure in Brazil before being acquired by insurance company Porto Seguro in 2021.
The founders believe businesses need to understand not only who is operating an AI agent but also what that agent is authorized to do.
Beltic also strengthened its identity infrastructure through the acquisition of Verifiet in 2025.
Verifiet was co-founded by Farhan Afsahi, who joined Beltic as chief technology officer following the acquisition.
According to Beltic, the acquisition added a data platform covering more than 500 million entities across multiple countries.
The company uses this data to help identify the people and businesses behind AI agents before evaluating what those agents are attempting to do.
Beltic is currently running pilots with companies as it develops its platform for AI-agent traffic.
The company’s nine-person team includes six engineers, its two founders and a lead focused on customer development.
The company says its goal is to help businesses safely accept legitimate AI-agent traffic while maintaining control over actions that could create fraud, security or transaction risks.
With its new funding, Beltic plans to continue developing its verification and authorization infrastructure as AI agents become more involved in online commerce and business workflows.
