Guests: Prasanna Venkatesh | Kartik Kumar
Company: Egen
Show Name: An Eye on AI
Topic: Agentic AI

Fraud detection is no longer a game of catching up. As fraudsters use AI to automate and scale their attacks, investigators are fighting back with the same weapon. Egen is helping public agencies turn artificial intelligence into a trusted assistant that surfaces hidden patterns, flags suspicious behavior, and lets human experts stay in control.

Fraud has always been a moving target. Every new technology that helps businesses or governments operate faster also opens new avenues for fraudsters. Prasanna Venkatesh, Sr. Engineering Manager, and Kartik Kumar, Sr. Data Engineer at Egen, are working to change how agencies respond to this challenge.

According to Venkatesh, the first problem is complexity. Public investigators often have to review massive volumes of claims — from unemployment benefits to identity verification. “Fraudsters are constantly innovating, using emerging technologies to find weaknesses,” he explained. “Investigators must stay one step ahead and look across multiple signals to confirm whether a claim is real — everything from email and employer data to wage information and even Social Security validation.”

Kartik Kumar added that much of this process is still manual. “Fraud detection today involves reviewing one claim at a time,” he said. “Fraud doesn’t usually happen in isolation — it happens at scale. Investigators don’t always have tools that can connect suspicious claims and detect coordinated activity.”

That’s where Egen’s approach comes in. The company has been developing AI-driven tools designed specifically for public-sector investigators, combining data analytics with visual dashboards to give them a complete picture of each case. “By analyzing historic claims data, we can identify patterns and assign a fraud score,” said Venkatesh. “That lets investigators prioritize the most suspicious claims while quickly clearing legitimate ones.”

The system also includes interactive dashboards that visualize emerging patterns. When multiple claims share the same phone number, IP address, or document pattern, the platform can flag those anomalies instantly. “We’re not just looking for one bad claim,” Kumar said. “We’re showing investigators clusters of related activity that could signal organized fraud.”

A major innovation is Egen’s Document Intelligence tool. Investigators often spend hours validating ID cards, W-2s, and utility bills. Egen’s solution automates much of that work. “It automatically cross-checks information across documents, looking for mismatched names, altered layouts, or suspicious edits,” explained Kumar. “For example, if the same driver’s license number appears on multiple claims or if a photo looks superimposed, our system can flag that.”

Venkatesh noted that the tool doesn’t just analyze documents in isolation. It also verifies them against other databases, such as DMV or Social Security records, giving investigators confidence in their decisions. “This cross-matching is extremely hard to do manually,” he said. “AI makes it possible to spot reuse of fake IDs or synthetic identities across multiple claims.”

But even with these advanced systems, Egen emphasizes that AI is not replacing investigators — it’s empowering them. “AI is just a tool,” Kumar said. “It’s helping adjudicators speed up their analysis while ensuring they remain in full control. Fraudsters are using AI to scale attacks, so investigators need AI to keep pace.”

Both experts see the fraud landscape evolving into what they describe as an “AI arms race.” Fraudsters are using generative tools to create synthetic identities and forged documents faster than humans can verify them. Agencies, in turn, are deploying AI to detect those same forgeries and connect cross-linked fraud rings. “The same technology used to create fake documents can also help us detect them,” said Venkatesh. “It’s about human intelligence and machine intelligence working together.”

Egen’s collaboration model with agencies is another key differentiator. Instead of offering a fixed product, the company customizes each solution based on the client’s data and workflow. “Think of it like a recipe,” Kumar explained. “Every agency has a different level of complexity. We work with them to fine-tune the model and dashboard configurations, just like adjusting spice levels in a dish.”

Venkatesh added that the close collaboration ensures not just technical fit but user adoption. “Fraud detection is a fast-changing space,” he said. “We need to constantly adapt the models and interfaces as new fraud patterns emerge and as investigators’ workflows evolve.”

Looking ahead, both leaders are optimistic about how AI and large language models will continue to reshape fraud prevention. Agentic AI and generative models can automate many repetitive tasks, from document review to cross-verification. “We’re moving from reactive to proactive,” said Kumar. “With agentic AI, investigators can get alerts before fraudulent claims scale up. The goal is to make fraud detection smarter, faster, and more adaptive.”

Venkatesh believes the future lies in true collaboration between human judgment and AI-driven insight. “It’s about the partnership between human intelligence and advanced AI,” he said. “That’s what will redefine fraud detection — not automation alone, but augmentation.”