Q&A: Shyam Bhatt, Quantexa

Let’s start with ‘Operation Pot Noodle’. What happened? 

The aptly named Operation Pot Noodle involved a network of apparently unconnected students making claims for stolen laptops or items which were being paid out. 

Viewed individually, the claims did not necessarily look remarkable. It was only when the insurer implemented our technology and saw the full picture that the relationships between the people, the claims, and the wider pattern became visible, which included phantom students, synthetic IDs and other previously unidentified risks. 

That is the important lesson. Fraudsters rarely present themselves as a convenient, clearly labelled network. They make small changes, spread activity across different identities and exploit the fact that insurers often assess claims one at a time. 

Operation Pot Noodle showed how connecting all available information accurately can turn a collection of seemingly routine claims into a coherent, contextual picture. 

Why is context so important in insurance? 

Insurance decisions are rarely difficult because insurers lack data. They are difficult because the relevant data is fragmented across policy, claims, billing, customer, supplier and third-party systems. 

A claim may appear legitimate when examined as an isolated transaction. Its significance can change when the insurer understands who is involved, whether those parties have appeared elsewhere, and how they are connected. 

The same principle applies beyond fraud. Underwriters need context around the customer and the risk. Claims teams need to understand the claimant, other parties and the supply chain. 

Customer teams need a reliable view across policies, products and channels. Context means understanding who somebody really is, how they are connected and why those connections matter to the decision being made in real-time. 

Take a commercial property claim as an example – the insurer was only provided with a contact and a business name. In a matter of seconds, through the unified view presented by Quantexa, they were able to identify the real director, financial stress, and shared links via contact information to other directors who were using Government Ministers’ addresses to give their companies a sense of legitimacy. 

The result was that a cannabis farm was identified with a claim worth over £250k. 

What prevents insurers from seeing that picture today? 

Most insurers have accumulated multiple systems through growth, acquisitions and successive technology programmes. 

Those systems may describe the same person, business or asset differently. Traditional matching can miss connections when information is incomplete, inconsistent or has changed. 

At the other extreme, overly broad rules can generate large numbers of false positives. Investigators then spend time assembling information manually rather than applying their judgement. 

The underlying challenge is therefore not simply finding more data. It is resolving fragmented records into trusted real-world entities and identifying the relationships relevant to a particular decision in real-time. 

During one implementation, an insurer identified £18m of missed fraud exposure before even going live. 

What does Quantexa aim to do differently? 

We combine dynamic entity resolution, network generation and contextual analytics. 

That allows an insurer to bring together internal and external information, identify the people, organisations, places and assets represented within it, and understand the relationships between them. 

Crucially, the result is explainable. Investigators and decision-makers can see why records have been connected and why a network or claim merits attention. 

The platform is also designed to complement existing systems rather than requiring insurers to abandon their established workflows. 

Context and analytical outputs can be made available through applications, APIs and data products, or embedded into platforms such as Guidewire ClaimCenter. 

One large tier 1 insurer implemented our technology as part of its group anti-fraud solution, leveraging existing investments and augmenting their existing capabilities rather than replacing them. 

This insurer saves £200m per year through this additional, complementary set of capabilities. 

Is this principally a counter-fraud capability? 

Fraud is an obvious starting point, because the value of relationships is so tangible. But the same connected data foundation can support claims segmentation, supplier intelligence, underwriting, customer intelligence, subrogation and litigation risk. That matters commercially. 

Insurers should not have to create a different version of the customer or risk for every use case. 

A trusted foundation can be reused across the business, with the context assembled according to the decision at hand. 

Typically, insurance fraud has been the proving ground rather than the destination, and insurers working with us have expanded into enterprise fraud detection, financial crime detection and even revenue generation solutions to increase growth, all using the same set of underlying platform capabilities.

Sponsored content