Veridox reimagines the first line of defence against insurance fraud

During ClaimsTech – The Pitch, Dan Sandler, Founder and CEO of Veridox – overall competition winner – outlined how insurers can improve both fraud detection and operational efficiency through a new approach to document intelligence.

Rather than relying solely on investigators manually reviewing thousands of documents, Veridox acts as a first line of defence, analysing every incoming file for risk before automatically allowing lower-risk cases to proceed and routing higher-risk files for human review. By ensuring every document is assessed rather than just a small proportion, insurers can identify more fraudulent activity while allowing investigators to focus their expertise on the highest-risk cases.

Sophisticated

Opening the session, Dan highlighted the growing challenge insurers face as fraud becomes more sophisticated and document volumes increase. He explained that the industry’s biggest constraint is no longer simply detecting fraud, but doing so consistently and at scale. Veridox addresses both challenges by combining multiple specialist AI capabilities to analyse documents, images and supporting evidence within seconds, presenting structured, explainable outputs that improve operational decision-making while increasing the likelihood of identifying suspicious activity.

According to Dan, the platform is designed to become an insurer’s first line of defence, helping organisations analyse every document consistently before it reaches an investigator, improving both fraud detection and operational efficiency.

Specialist AI

Demonstrating the platform, Dan showed how users can upload individual documents, images or complete case files through a web-based dashboard. Once submitted, each file is analysed using multiple specialist AI modules, including optical character recognition, QR code analysis, font analysis and external verification, before the results are combined into a single, evidence-backed assessment.

Rather than producing a single risk score, Veridox presents explainable outputs, beginning with headline observations before progressing to detailed technical findings, supporting evidence and recommended next steps. Every conclusion is backed by a transparent audit trail, enabling investigators to understand how the platform reached its findings before making a final decision.

Transparency

The presentation emphasised that transparency is central to the platform’s approach. Rather than simply assigning a risk score, Dan said the system is designed to provide investigators with the evidence supporting each finding, enabling them to review, challenge or accept the conclusions before progressing an investigation.

Delegates were also shown how investigators can interact directly with the platform through an integrated AI assistant, allowing them to explore evidence further, generate summaries, prepare supporting documentation and ask follow-up questions without leaving the case file.

Where investigators disagree with the platform’s assessment, Dan explained that they can amend the suggested risk profile and record the reasons for doing so. He said those decisions help refine future outputs within the organisation’s own implementation of the platform, allowing the technology to evolve alongside operational requirements.

Although the presentation centred on insurance claims, Dan suggested insurers are increasingly exploring wider applications for the technology. He explained that some organisations are considering its use across underwriting, policy validation, employee compliance and other operational processes where document verification forms part of decision-making.

AI Courtroom

The session then turned to one of the industry’s fastest-growing challenges: distinguishing authentic from AI-generated images. Dan explained that as synthetic content becomes increasingly convincing, insurers need better ways to assess authenticity while understanding the reasoning behind any conclusion.

He demonstrated Veridox’s AI Courtroom, where separate AI defence and prosecution agents each build a case for and against an image being genuine. Their arguments are then presented to an independent AI judge, which weighs the evidence before delivering a verdict and written judgement.

Rather than simply labelling an image as real or fake, the feature provides investigators with a clear explanation of the evidence considered and the reasoning behind the decision, supporting confident, transparent decision-making.

Evidence based

Reflecting on his background as a solicitor, Dan explained that evidence remains at the heart of every successful fraud investigation. While AI can help identify risk and surface information far more quickly than manual review alone, he argued that investigators must always be able to understand, explain and justify the conclusions they reach – particularly where claims are challenged or progress to litigation.

Dan also spoke about Veridox’s product philosophy, explaining that the platform is developed in close collaboration with insurers. Customer feedback, he said, plays a central role in shaping new capabilities, ensuring the platform continues to evolve alongside changing fraud trends and operational requirements.

Accuracy remained a recurring theme throughout the session. Dan argued that the real value isn’t simply identifying more fraud, but providing investigators with the evidence and context needed to make faster, more confident and more defensible decisions.

Human decision making

Despite the rapid advancement of AI, Dan was clear that investigators remain central to the process. Veridox, he said, is designed to strengthen human decision-making, not replace it, acting as an intelligent first line of defence that allows experienced investigators to focus their time where it adds the greatest value.

Closing the presentation, Dan said the fight against fraud will continue to evolve as AI becomes more sophisticated. Success, he suggested, will depend on insurers and technology providers working together to build faster, more consistent and more effective fraud investigations.

Five finalists – Fixzy AI, FraudOps, Nolana, SwiftCase and Veridox – made it through to the final of ClaimsTech – The Pitch 2026 following qualification stages including a newly introduced, online semi-final programme. The event was hosted by headline sponsor Kennedys and supported by Corporate Partners Solera Audatex and Verisk. Sponsors HLBS, Shift and Unitary also supported the programme, alongside Associate Partners Finpro, InsurLab Germany and Insurtech UK.