The security process can make or break an airport. When the queues are fast-moving and the staff are friendly, holidaymakers feel relaxed. The smooth but thorough process helps nervous travellers feel more reassured, too. But when you have hour-long waits, stressed-out security staf,f and machines that seem to beep at every bag… the experience quickly goes downhill. 

It's the same story for KYC and customer onboarding. Everyone needs to go through onboarding, just like everyone needs to go through airport security. But a bad experience can lead to customer frustration and compliance stress. More than nine in ten companies lose customers during the onboarding stage too, seriously hurting profit margins. 

In this blog, we're running with the airport security analogy to explain how reliable technology, like Sardine's AI agent for fraud preventions - smoothens the process without compromising safety. In our experience, Sardine's AI agents cut 75% of routine compliance casework, making for much happier compliance offers. AI agents help across other areas of the firm too, creating efficiencies all the way through front-to-back offices. 

1. AI Fraud Detection: Pre-checks before onboarding even begins 

This step gathers the data our agents will need to create a first class experience.

To even enter the airport security zone, travellers need to scan their boarding pass. And to get a ticket in the first place, flyers need to provide identification such as their passport. This first check helps to keep out unauthorized people, fakes or criminals. 

This mirrors KYC or any customer onboarding. During the sales and marketing stages, the Sardine AI platform collects data, conducts initial screening and scanning information against vital risk indicators. They run deep analysis on the user’s geolocation, device information, mouse movements, IP address and over 4,000 more factors… without causing any friction to the customer. 

When a bad actor is detected, such as a gen-AI bot, they are blocked from proceeding any further. The intelligence can also be stored and shared with a global network like Sonar, to further strengthen our shield against fraud. 

2 .Baggage check” for financial history and Compliance Monitoring

Around 2 million accounts exhibit signs of money mule behaviour. Finding patterns between them is essential for preventing more being created. 

During the onboarding stage, Sardine's platform uses 3rd party data, telco data, and, bank consortia check other accounts in the customer's name. It scans financial history or "financial baggage” for potential blacklisting.  If there are signs of bad financial standing, or suspicious activity, the platform often creates an alert for the compliance office.

This is where Sardine's Agentic AI steps in. 

Similar to a bag check, where travellers place their hand luggage into trays to be scanned by machines, and overseen by staff. Any suspicious activity picked up by the platform is passed over the AI agent for closer scrutiny - you could think of the Sardine agent like the security officer that takes you aside. But instead of waiting and biting their nails for what seems like hours, the extra check is instantaneous, and customers will hardly feel it. 

The idea of doing all of this manually for hundreds of thousands of people seems ludicrous now. But the reality is that for years financial institutions have been manually checking documents. This has led to inconsistencies, errors, oversights and long backlogs. Even today, with increased 95%+ of compliance officer time is spent on false positives.

AI agents have demonstrated 100% precision in detecting false positives during our testing. This frees up compliance officers to work more compliex cases. In turn, putting compliance offers into oversight roles. This is a much better use of their skills and capabilities. 

3. 360-degree body scan for "Step Up" KYC or KYB

If the customer is "stepped up" for manual review, financial institutions or businesses will often look at customer documents, selfies or 3rd party data. This is difficult for traditional automation because there can be 1,000s of possible edge cases, like cultural variations, nicknames, stolen credentials or foreign languages

A 2024 report from IBM found that breaches involving stolen or compromised credentials take the longest to identify and contain, at 292 days on average. This can create chaotic working environments. Skilled officers are forced to react under pressure, as they try to put out fires and avoid fines, taking time away from proactively blocking new fake account attempts. 

Sardine's AI agents help compliance teams to take back control. As the customer is onboarding, the identity documents are checked against video facial recognition, as well as the device itself. The AI Agent then produces a full rationale based on these inputs and provides that to a compliance officer. This can compress the time "stuck in a queue" from ~days to minutes. One Sardine client cleared their 20 day backlog of manual reviews by having our AI Agents in production for a month.

If there is a suspicion, the agents are as adept as humans, and in many cases more so at determining whether it is true positive or a false positive. This saves significant workloads for compliance officers. One of our partners, Raise, cut down the time spent on manual reviews by 98% by adopting this technology. 

In the same way that an airport body scanner can detect a prohibited item on a traveler better than a pat-down, AI agents are usually much more effective and comfortable for the user as well. 

4. “Who's bag is this?” employees take over for specialist cases: Fraud Protection Expert 

Humans are always accountable when AI Agents are used in production. In a minority of cases, the airport security machines alert employees about a problem that needs to be resolved. This could be as innocent as an accidental aerosol can, water bottle or pair of scissors within the luggage. Or it could be something more sinister. In this scenario, trained airport staff take the traveller aside and go through the issue in detail. 

This is how Sardine's AI agents help compliance officers too. By providing a summary, compliance officers can take a closer look themselves and either approve or reject the AI Agent's decisions. Compliance officers have all the data, information and regulations at their fingertips, with easy-to-use dashboards. 

‍

5. Constant checks and monitoring even after onboarding: Real Time Transaction Monitoring 

Step Up authentication can and often does happen after a customer has onboarded if a customer demonstrates enough risk. After customers have been onboarded, the constant checks and scans by Sardine's AI continue. This is where the airport analogy ends. To picture the true extent of Sardine's capabilities, you would have to imagine that the bags and travellers are being constantly scanned 24/7 across thousands of different metrics, regulations and sanctions lists. After they land, in the hotel, on the trip home and beyond.

First pass findings are continuously checked, double checked and checked again. Recommendations and insights are sent to relevant personnel across the front, middle and back offices too. If anything changes with the customer, for example, a login from a new location, device or language, Sardine's AI agents flag the instance the moment it happens and block where necessary. 

Fintech Fraud Prevention: Better security starts here 

Security is a vital part of running a fintech service. Not just for the protection of other customers, but also for the wellbeing of employees, and the longevity of the company.

The average cost of a data breach rose to $4.9 million in 2024, from $4.5 million in 2023, and it's growing all the time. As we saw from high-profile cases last year, some are even forced to close down in disgrace. 

It's little surprise that two out of three organisations are now deploying AI and automation across their operations, a 10% jump from last year. But the real question is, which is the best one on the market? 

Today, all airports have some form of security, but some are clearly light years ahead of others. The experience you have with excellent machines with improved features (for example, when you don't need to remove liquids or laptops from hand luggage) is clearly more reassuring than those which take hours and create constant false alarms. 

If you'd like to support your employees with a first class team of ai fraud detectionAI agents and turbo-charge security, we'd love to demonstrate how we can support. Together, we can fly.