The Pulse
Beltic Raises $7.3 Million to Verify AI Agents
Beltic is emerging from stealth with a $7.3 million seed round led by Norwest, bringing its total funding to $8.8 million. The San Francisco startup says its platform checks an AI agent’s identity and authority before it acts for a person o
AI.info Team ·
“For decades, a large part of risk management was asking and verifying, ‘are you human?’”
Isha Bhatnagar, CEO and co-founder, Beltic
Beltic raises $7.3 million for agent checks
Beltic emerged from stealth on September 30 with a $7.3 million seed round led by Norwest, saying it is building identity and authorization checks for AI agents that act on behalf of people and businesses. Restive Ventures, Oxford Seed Fund and Collide Capital also participated. The company says its total funding is $8.8 million, including a $1.5 million pre-seed round led by Collide Capital.
The San Francisco company’s pitch is that existing fraud and identity systems are built around human users, while autonomous software can initiate actions without a person present at each step. Beltic says its platform verifies both an agent and a transaction, so a business can assess who is behind the software and whether it is permitted to take a particular action.
‘Know your agent’ extends checks beyond the customer
Beltic calls the approach “know your agent,” or KYA, adapting the financial industry’s “know your customer” model to software acting for a customer. The company describes the checks as identifying an agent, verifying the person or organization behind it, and determining whether its actions are authorized in real time. It says the platform is intended for businesses, payment providers, financial institutions and card networks.
Bhatnagar’s comparison to human identity checks points to a practical problem for merchants: a request may come from software, but the business still needs to establish whose agent it is and what that agent may do. The company also argues that verification must follow delegation between agents. If one agent hands work to another, Beltic says, the receiving system should not simply inherit trust without checking the new handoff.
“Most people envision a simple chain, where a human authorizes an agent and that agent acts, but the reality is recursive,” Bhatnagar said in the announcement. “Agents call other agents, orchestrate further agents beneath them, and hand work to systems they do not own.”
Founders bring identity and payments experience
Beltic says Bhatnagar spent six years at Coinbase building identity and onboarding infrastructure for more than 50 million users and the company’s first institutional clients. Co-founder Mike Allan previously founded Atar, a banking-as-a-service provider that Beltic says scaled across Brazil’s payments and core banking systems before its acquisition by Porto Seguro. CTO Farhan Afsahi co-founded Verifiet, which Beltic acquired in 2025.
The company says that acquisition gave it an entity-data platform covering more than 500 million entities across jurisdictions. That background underpins Beltic’s argument that agent verification depends not only on identifying software but on checking the person or business responsible for it. The announcement does not provide independent performance measurements or describe how the platform handles disputed or mistaken authorizations.
Early access goes to selected companies
Beltic says it is working with a select group of companies in agentic commerce and payments. Those design partners will receive early access and help shape standards for agent transactions, according to the announcement. The company invites other interested businesses to contact its team, but does not name the partners or give a broader product launch date.
The release leaves key commercial details unspecified, including pricing and the names of current customers. For now, Beltic’s stated offer is a verification layer that checks the agent, its principal and its authority before a transaction proceeds.