Adi
Goel

The year was 2017. Adi Goel was a vice president at Deutsche Börse, Europe's largest exchange, managing a $200 million venture fund and writing Medium articles invoking Tony Stark's AI suit as a metaphor for the future of asset management. Most people at that stage of a finance career double down on the finance. Adi went to Revolut instead.

That pivot tells you something. He joined the British neobank as Head of Product and Operations for the US and New Markets - which at the time meant: US market, zero customers, no infrastructure, a blank page. When he left two years later, Revolut had millions of US customers. He had built the product stack, the compliance plumbing, the banking partnerships. He had also watched, up close, how broken the fraud and compliance tooling was for any fast-growing fintech trying to operate at scale.

So in February 2020, alongside two Revolut colleagues - Soups Ranjan, who had previously been Head of Data Science at Coinbase, and Zahid Shaikh, the company's Chief Product Officer - Adi co-founded Sardine. The founding insight was blunt: financial crime prevention tools were fragmented, siloed, and built for a different era. Every company was buying ten vendors, stitching them together with duct tape, and still losing to sophisticated fraud rings. Sardine would be the single API.

Five years later, that API has screened more than $1.36 trillion in payments, fingerprinted 5.75 billion devices, and is trusted by over 300 enterprises across more than 70 countries. The customer list spans neobanks like Brex and Chipper Cash, crypto exchanges including Luno and MoonPay, and the kind of large financial institutions that used to handle compliance with armies of analysts and legacy COBOL systems.

Sardine sits at the intersection of three disciplines that are usually handled by different teams with different tools: fraud detection, AML compliance, and identity verification. The company's platform combines behavioral biometrics - how you hold your phone, how fast you type, the rhythm of your mouse movements - with device intelligence, transaction monitoring, and AI-driven workflow automation.

The behavioral piece is the strangest and most powerful part. Before you ever tap "send" on a payment, Sardine's SDK has been watching the session - not your personal data, but the behavioral fingerprint of how you interact with an app. A legitimate customer and a fraudster using stolen credentials interact with an interface in measurably different ways. That signal, layered with device intelligence and transaction history, gives financial institutions a real-time risk score that updates as the session unfolds.

In 2024, Sardine launched a suite of AI agents - autonomous systems that handle investigative workflows end-to-end. The KYC Agent achieves an 88% auto-resolution rate. A tier-1 bank using Sardine prevented 42% of wire fraud. A commercial neobank cut chargeback losses by 70%. These are not marketing claims from a pitch deck - they are the numbers Adi cites in interviews with an engineer's precision.

Adi Goel studied electronics engineering at IIT Delhi - one of India's most competitive technical institutions - with concentrations in artificial intelligence, operating systems, and signals processing. Then Wharton for an MBA in finance and statistics, where he became vice president of the Wharton FinTech club and co-founded the Penn Artificial Intelligence Society. Before startups, there was PIMCO, trading fixed income securities at one of the world's largest asset managers.

That path - engineering precision, quant finance fluency, product leadership, operations at scale - is the exact combination that Sardine needs to exist. Fraud prevention at the enterprise level requires someone who can speak to a CISO about behavioral anomaly detection, to a CFO about return on compliance spend, and to an engineering team about API architecture. Adi does all three.

In his 2017 Wharton FinTech article - titled "Why We Need Tony Stark's Suit" - he was arguing that machine learning would fundamentally transform how humans and financial systems interact. He wasn't writing about fraud detection then. But he was already thinking about the same problem: how to take human judgment, encode it in a system, and make it run at machine speed.

The Tony Stark metaphor - a human amplified by AI, not replaced - is exactly what Sardine's agentic platform looks like in 2025. The KYC agent doesn't eliminate compliance analysts; it handles the 88% of routine cases automatically so analysts can focus on the 12% that actually require judgment. Three years before "agentic AI" became a buzzword, Adi had already articulated the model.