A San Francisco startup you've likely never heard of is betting that artificial intelligence needs its own breed of bouncer—one that can verify whether a bot has permission to spend your money.
Beltic emerged from stealth on September 30, 2026, announcing it raised $7.3 million in seed funding led by Norwest. Restive Ventures, Oxford Seed Fund and Collide Capital joined the round, bringing the company's total haul to $8.8 million including an earlier $1.45 million raise, according to company disclosures. The pitch: as AI agents begin executing financial transactions autonomously, someone needs to verify they have the authority to do so.
It's a specific problem, perhaps narrower than most fledgling startups would choose. But CEO Isha Bhatnagar, who previously led product for global regulatory experience at Coinbase, believes the window is opening fast. Traditional identity verification asks "are you human?" she noted in a statement. The infrastructure Beltic is building asks something different—whether an autonomous agent conducting a purchase or fund transfer is acting within its prescribed boundaries.
The platform verifies three layers for each transaction: identity and origin, authorization through policy rules, and what the company describes as intent-based risk evaluation. Beltic claims it delivers results in under a second, a timeline Bhatnagar reiterated in a September interview with This Week in Fintech. Speed matters when transactions flow in real time and any friction costs conversions.
Beltic's approach got a boost in December 2025 when it acquired Verifiet, a business-identity data platform that brought coverage of more than 500 million entities, according to a Norwest blog post detailing the investment rationale. Verifiet co-founder Farhan Afsahi joined as CTO through that deal. The other co-founder, Mike Allan Pellin, previously built Atar, a Brazilian banking-as-a-service firm that Porto Seguro acquired in 2021.
The team remains lean. Nine people total—six engineers, the two co-founders and a go-to-market lead, per This Week in Fintech's reporting from late September. Advisors include Melissa Strait, Coinbase's former chief compliance officer, and Max Freeman, an SVP at Ramp. That roster suggests the company is threading a needle between compliance-heavy financial infrastructure and the developer-friendly world of emerging fintech tools.
Beltic incorporated in Delaware on June 18, 2025 and filed its Form D with the SEC on September 1, 2026, according to state records and securities filings. The company lists offices on Market Street in San Francisco, though like many startups in its cohort, it's hiring remotely across a wide swath of time zones—UTC-5 to UTC+3, spanning the Americas through parts of Europe and Africa. Open roles include another go-to-market hire and a machine learning engineer.
The customer base Beltic is targeting runs the gamut: merchants, payment providers, financial institutions and card networks. The company says it's working with what it calls "global commerce giants" as design partners but declined to name them publicly, a familiar dance for early-stage startups anxious not to spook potential clients or oversell traction.
Jordan Leites, a principal at Norwest, framed the investment in broader terms. "As money moves increasingly in real time, identity and trust infrastructure has to keep pace," he said in the company's announcement. Norwest's team positioned Beltic at the intersection of stablecoin infrastructure and what some are calling "agentic commerce"—a term that hasn't quite settled into common usage but captures the idea of AI systems executing transactions with limited human oversight.
Whether that intersection represents a durable market remains to be seen. The concept of "Know Your Agent" verification is gaining traction in compliance circles, with companies like Experian, Trulioo and Sumsub all exploring variations on the theme over the past year. That simultaneous movement suggests genuine demand, or at least investor conviction that demand is coming. It also means Beltic will need to move quickly to establish itself before the category gets crowded or larger incumbents decide the opportunity justifies building in-house.
For now, the company is making the standard early-stage bet: that defining a new category before others arrive carries advantages that outweigh the risk of being too early. Whether AI agents will conduct enough financial transactions to sustain a standalone verification layer is the multi-million-dollar question Bhatnagar and her team are wagering they've answered correctly.
