Sardine

Deepfakes aren't a buzzword. They're a reality. They've taken high impact scams and masking attacks to another level. We're starting to see this challenge reach critical mass: 👉 AI-generated scam content will become increasingly sophisticated 👉 Social engineering attacks will leverage deepfake technology for voice and video 👉 Scams that start in social media, then move to crypto and banking exploit the gaps between companies, networks and industries. 👉 First-party fraud will continue its explosive growth 👉 The intersection of these trends creates a perfect storm. Financial institutions can't simply write off these losses - the numbers are too large. They can't just add friction - the competition is too fierce. They need a new approach entirely. What we see our strongest customers doing is doubling down on ML/AI and training custom models to pick up on unique risk flags. They're also participating in cross-industry consortiums to close up key data gaps. And they're making heavy investments into AI to beef up their risk operations.

How to use AI Agents in Production in financial services. Imagine they’re human agents but faster. Here’s the process. 1. Train them on your procedures or workflows 2. Give them access to the right systems and data to do their job (securely) 3. Make sure they provide evidence for their decision 4. Make sure there's an audit log of everything 5. Make sure its all reported up to any risk comittees 6. And finally, make sure you can explain what happened to a regulator This metaphor works for AI Agents, the process just looks a little different in practice. We’ve been using AI agents in production for highly regulated use cases for the past three months. We’ve found they produce incredible results. Drop a comment below if you want a link to the full whitepaper 👇

Welcome to Sardine[Con] - Ending the Scamdemic: Our 1 day event in SF on the 20th August 20205. No more talk. It’s time to take action. We’ve put together a line up of Erin West, Dr Nicola Harding, and of course Soups Ranjan, and Matt Vega, 3CI, SIP, ECFE, CPFPP. We’ll share global best practices on dealing with scams. But then, we need you. Bring your ideas, your ambitions, and your commitment to reducing scams. Let’s do this. Mark the 20th of August as a date for your diaries. The event is on the 20th August 2025 at the Pearl in San Francisco Head to Sardine.ai/sardine-con to register.

Can't think of the rule to catch that spike in scams? Try building one with AI. When you know what you want to do but you don't know quite how to build it. That's when AI is at its best. For example "Detect if a user shares a device with another customer who has a non-US IP" (pictured below) Say a rule in natural language and Sardine turns it into a platform rule. Want to give it a shot? Come for a demo :)

In real world testing, our KYC Agent correctly resolved 100% of false positives that traditionally plague compliance workflows. Our KYC agent + platform transformed a 14-hours of daily work into just 45 minutes for one client. For BSA/AML purposes, you only need a verified SSN to onboard low risk users. But in cases where synthetic or stolen identity fraud is suspected, we ‘step up’ to a documentary KYC - passport, driver license with a selfie. The problem? Up to 95% of these are false positives. This means good users are stuck in a queue, and less likely to convert. Our AI Agent works those "manual review" queues and compressed a clients manual review backlog from 20 days to 2 minutes. Here’s how it works. - We train an agent CIP (Customer Identification Program) process followed by the compliance officer. - The agent evaluates matches using context from multiple data sources like Doc KYC information in the Sardine platform - Automatically identifying false positives like name matches with mismatched birthdates. Our AI can handle - Cultural name variations, - Different date formats, - and complex document verification sequences The main goal is to free up compliance officers from the mind numbing work of false positives (and the associated burn out) to focus on complex investigations. KYC doesn't have to be a bottleneck. Have you tried this in your onboarding flow?

🎁 Feature Drop: Our AI now AI transforms anomalies into rules, automatically Now you can go from - New fraud pattern - New rules suggested - Quick back test - To live in production Faster than ever before. Here's how it works - Our Anomaly Engine monitors your traffic for anomalies - Then it triggers our rule suggested - Which suggests new rules for you to create to catch this pattern Because our models are explainable, we know what features contributed to the anomaly. Because we can build rules with our AI rule suggester, we can build rules that catch new patterns. We're beginning to reap the compounding benefits of - ML - AI based rule generation - and a platform that’s real-time for testing and creating rules Every day we’re trying to help you move faster than the threat.

Our team built an AI that can detect fake job applications. Here’s how it works 👇 Couriers, e-commerce and tech companies see so many fake or suspicious job applications. Users fake their location (with proxies or VPNs), and even video interviews using deep fakes. Just last week, we had a candidate who reached the offer stage of our interview but then our head of people had a Spidey sense, and suggested for them to meet up in person. The candidate completely balked and deleted his LinkedIn profile! There are nefarious bad actors ranging from - Getting a higher US salary while living in a lower cost region by faking their location - Criminal groups getting jobs as couriers or delivery to steal or misdirect deliveries - Criminals or bad actors working in delivery warehouses to misdirect or steal goods - Bad actors sponsored by rogue nations looking to get software engineering jobs for IP theft or vulnerability exploits Our solution? By embedding Sardine’s fraud capabilities into Vertical HR systems (like Ashby, Zoho, Rippling etc) we can find out - The candidates true location - By identifying if they’re using a VPN or proxy - If the candidate is face swapping during the video stream - By profiling the video call on device The best part? This came from a hackathon. It's amazing what the team comes up with when they get to solve problems with our product. The idea came from our head of people Chandni, and our Head of Engineering Kazuki walks through how it works in the video below.

How to use AI Agents in Production in financial services. Imagine they’re human agents but faster. Here’s the process. 1. Train them on your procedures or workflows 2. Give them access to the right systems and data to do their job (securely) 3. Make sure they provide evidence for their decision 4. Make sure there's an audit log of everything 5. Make sure its all reported up to any risk comittees 6. And finally, make sure you can explain what happened to a regulator This metaphor works for AI Agents, the process just looks a little different in practice. We’ve been using AI agents in production for highly regulated use cases for the past three months. We’ve found they produce incredible results. Drop a comment below if you want a link to the full whitepaper 👇

Welcome to Sardine[Con] - Ending the Scamdemic: Our 1 day event in SF on the 20th August 20205. No more talk. It’s time to take action. We’ve put together a line up of Erin West, Dr Nicola Harding, and of course Soups Ranjan, and Matt Vega, 3CI, SIP, ECFE, CPFPP. We’ll share global best practices on dealing with scams. But then, we need you. Bring your ideas, your ambitions, and your commitment to reducing scams. Let’s do this. Mark the 20th of August as a date for your diaries. The event is on the 20th August 2025 at the Pearl in San Francisco Head to Sardine.ai/sardine-con to register.

Our AI agents, in production, cut customer KYC wait times from 20 days to 2 minutes. 95% of BSA/AML alerts are false positives. 85% of corporations hate bank onboarding. Major banks average 307 employees just for KYC alone. This costs customers and revenue. We deployed AI agents in production environments at multiple financial institutions for 3+ months and show AI Agents can meaningfully improve KPIs Our new whitepaper breaks down the key findings and key results like: KYC backlogs: 14 hours → 41 minutes Customer wait time: 20 days → 2 minutes Resolution accuracy: AI agents follow operating procedures in 100% of cases vs <95% for humans FI's rightly worry, what about hallucination? What about data privacy? Will the regulator allow it? These live, production data points are all within existing regulatory frameworks (SR 11-7 compliant). Our Agentic Oversight Framework maintains complete human accountability while delivering: - Alignment to Standard Operating Procedures (SoPs) - A full audit trail of every data element accessed - A full, explained decision rationale, reviewed before every case is progressed - Continuous learning from expert reviewers - Automated drift detection and safeguards Read the full white paper: The white paper is a playbook for how financial institutions can safely implement agentic AI while fully complying with regulatory requirements. Real results. Real institutions. Real transformation.