San Francisco-built AI-native fraud, AML and credit-risk platform for global fintech — founded 2020 by ex-Coinbase / Revolut / Uber risk leaders, $145M raised, used in 150+ countries
Sardine is one of the most credible AI-native fraud-prevention and compliance platforms globally — founded in 2020 in San Francisco by Soups Ranjan (CEO, formerly Director of Data Science & Risk at Coinbase), Aditya Goel (ex-Revolut, ex-Uber), and Zahid Shaikh (ex-PayPal). Sardine raised a $70M Series C in February 2025 led by Activant Capital with Andreessen Horowitz, Google Ventures, Nyca, Geodesic, Moody's Analytics, Experian Ventures and others — bringing total funding to ~$145M. The product combines device intelligence, behavioural biometrics, and machine learning into a real-time fraud / AML / credit-risk decisioning surface, and 300+ enterprise customers worldwide use it including FIS, Ascensus, Deel, GoDaddy, X (Twitter) and a long tail of crypto exchanges and neobanks. For Indian product teams, the right framing is: Sardine is the right tool when your fraud problem has global, cross-border or crypto-exposed shape — Indian fintechs serving offshore customers, crypto exchanges, B2B cross-border payments. For India-domestic fraud problems built around Aadhaar, UPI and CIBIL, you'll typically pair an Indian-stack vendor (HyperVerge for video KYC, IDfy or Karza for identity-data, Sift / Forter for general fraud) with Sardine layered on top of the global / crypto surface — or skip Sardine entirely for purely India-domestic flows.
Sardine is an AI-native fraud-prevention, AML / sanctions-compliance and credit-risk platform aimed at fintechs, neobanks, payment processors, marketplaces and crypto businesses. The product surface is a real-time risk-decisioning layer that ingests a rich set of signals — device fingerprint, behavioural biometrics (typing rhythm, mouse movement, gesture patterns), IP & geo, transaction context, identity-document signals, plus consortium data shared across customers — and runs them through Sardine's machine-learning models to score each event for fraud, money-laundering or credit risk, in milliseconds, with a configurable rules layer on top.
The company was founded in 2020 in San Francisco by three founders with deep risk-and-compliance backgrounds: Soups Ranjan (CEO) was previously Coinbase's Director of Data Science and Risk, where he built much of Coinbase's early fraud-detection infrastructure. Aditya Goel (Head of Payment Products) came from Revolut and Uber, and Zahid Shaikh (Head of Risk Products) from PayPal. The founding-team thesis was simple: traditional fraud platforms (LexisNexis, Experian, FICO) were built for a banking world where most fraud had a paper trail; the new world — instant onboarding, crypto, cross-border B2B — needed a real-time, ML-first risk platform that didn't depend on bureau pulls.
Funding history: Seed in 2020, Series A in 2021, $51.5M Series B in September 2022 led by Andreessen Horowitz, and a $70M Series C in February 2025 led by Activant Capital — bringing total funding to roughly $145M. The Series C explicitly funded an investment in agentic AI — Sardine is building autonomous-agent workflows that can review fraud cases, run KYC investigations and execute compliance tasks with reduced human-analyst time. This is the company's strategic direction through 2026.
For Indian product teams, the most important positioning point is that Sardine is fundamentally a global fraud platform, with depth in cross-border and crypto risk. It is widely used by Indian-founded global fintechs (cross-border B2B payment companies, Indian crypto exchanges that serve global customers, US-incorporated Indian-founder fintechs targeting US end-users). It is less commonly used as the primary fraud platform for India-domestic UPI / Aadhaar-based fintechs — for that, the coverage of Indian-specific data sources (CIBIL, Aadhaar mismatch signals, UPI VPA reputation) is genuinely better at Indian-built vendors.
Typing rhythm, mouse movement, scroll patterns, touch gestures, copy-paste behaviour. Used to distinguish human users from bots, and known users from account-takeover attempts. Among the deepest behavioural signal stacks in commercial software.
Device fingerprinting, browser signals, emulator / virtual-machine detection, jailbreak / root detection, app-tampering signals. Real-time signal generation at the page-render or app-launch step; integrates via JavaScript SDK or native mobile SDKs.
ML-scored decisions on application, login, payment and high-risk action events — sub-100ms latency. Configurable rules layer for compliance teams to override or augment ML scores. Supports A/B testing of risk rules, which is unusual in this category.
Real-time screening against OFAC, EU, UN, India RBI, and other sanctions lists, plus PEP screening, adverse-media screening and crypto-wallet risk scoring. Case-management workflow for compliance team review.
Wallet-address risk scoring, on-chain transaction analysis, mixer / tornado-cash exposure, and integration with Chainalysis / TRM Labs. The deepest crypto-fraud surface among general fraud platforms — driven by the team's Coinbase background.
Newer surface: autonomous-agent workflows that review fraud cases, conduct KYC investigations, draft Suspicious Activity Reports (SARs) and execute escalation logic. The flagship investment area for the post-Series C product roadmap.
Sardine does not publish list prices. Real-world contracts in 2025–2026 typically land in the $30K–$200K+ per year range depending on transaction volume, which surfaces (fraud only vs fraud + AML + crypto) you license, and whether the Sardine team handles the implementation. For Indian buyers that translates to roughly ₹25L–₹1.7Cr+ per year all-in, with USD invoicing and 18% IGST. The page's previous "₹3,00,000+/year typical" figure (~$3.5K) was an order of magnitude too low for actual enterprise contracts; we've corrected it. Sardine also charges per-API-call fees on top of the platform fee for high-volume customers — typically a few cents per scored event. Always validate via a quote on sardine.ai.
Sardine is the wrong call when: you're a domestic Indian fintech where most of your fraud is rooted in Aadhaar/UPI/CIBIL data — use HyperVerge, IDfy or Karza as the primary stack and only consider Sardine as a behavioural-biometrics layer on top. Sardine is also the wrong call for very early-stage startups that need a fraud tool live this week — sales cycles are enterprise-paced.