From Documents to Real ROI: Live @ ITC with Hyperscience

Andrew Vogeney (Ideas Man):
Hi everyone, we’re here on the final day of ITC, live on the Monitor stage. I’m joined by Chip VonBurg, Field CTO at Hyperscience. Chip, how are you doing?

Chip VonBurg (Hyperscience):
I’m doing great. Thanks for having me. I’m excited to talk about the challenges insurers are facing and how technology and AI can help. It’s been the theme of almost every conversation here this week.

The Biggest Challenge in Insurance Today

Andrew:
In your opinion, what’s the biggest challenge in insurance right now?

Chip:
It’s a few things. First, rising customer expectations. We live in a Netflix-and-Amazon world where you get instant access to what you want. People expect that same immediacy from their insurers.

But insurance is still heavily document-based. Whether it’s a new policy, an accident, a claim, or a renewal, everything revolves around documents. Balancing real-time customer experiences with manual, paper-heavy processes is incredibly hard.

Add legacy systems on top of that—rigid, non-nimble workflows, disconnected data, and slow processes—and it becomes even more difficult to meet modern expectations.

The “PDF Problem” and Why It Still Matters

Andrew:
A lot of people say, “We did a digital transformation; now everything is PDFs.” Does that solve the problem?

Chip:
Not really. PDFs are often just digital paper. Most insurers still process PDFs the same way they processed paper. A McKinsey report from earlier this year found that 70–80% of insurance automation use cases are still document-related. The industry still fundamentally runs on documents and static formats.

Where Hyperscience Fits In

Andrew:
So where does Hyperscience come in?

Chip:
Hyperscience is an AI software company and a leader in the Intelligent Document Processing (IDP) space. We use ML and AI to read, understand, and structure data from complex documents—claims, accident reports, forms, you name it.

Our goal is to turn all of that unstructured incoming information into clean, usable data that can power back-office operations and fuel AI initiatives, including LLMs and RAG systems. Better data means faster processes, better accuracy, and a stronger foundation for automation and analytics.

Real Customer Examples

Andrew:
Give me one or two use cases where customers are seeing big gains.

Chip:
A great example is Corebridge Financial. They retired a legacy account-opening system in their life and retirement business and replaced it with Hyperscience. The results:

• 70% reduction in data entry costs
• Handwriting accuracy increased from 10% to 95%
• Over 1,000% ROI, because they process tens of millions of documents

Another example is the U.S. Department of Veterans Affairs (VA). They process 1 billion claims a year with Hyperscience, resulting in $470 million in annual labor savings. But the most important impact is on veterans: claims that used to take weeks are now typically resolved within the same day.

Why Clean Data Unlocks Efficiency (and AI)

Andrew:
Once you extract all this data, how does it actually unlock efficiency?

Chip:
Most people come to us for speed and cost savings—the obvious benefits. But the real unlock is gaining access to structured, reliable data that was previously trapped in documents.

Clean data improves everything: trend analysis, fraud detection, claims processing, and especially AI initiatives. Data is the fuel for AI. If your data isn’t good, your AI outcomes won’t be either.

Hyperscience can consume and feed RAG-based architectures (retrieval augmented generation). RAG lets companies query their own internal data—not just public sources—so they can make highly accurate, context-rich decisions.

Why So Many AI Projects Fail

Andrew:
There was an MIT study saying 95% of AI pilots fail. Why?

Chip:
Many companies aren’t AI-first. They try to bolt AI onto legacy processes or use it in areas where it can’t deliver real ROI. Without the right data foundation, AI initiatives stall.

Hyperscience customers, on the other hand, achieve automation rates of 98% at 99–99.5% accuracy because the data going into their systems is reliable and consistent. That’s what makes AI projects succeed.

The Future of AI in Insurance

Andrew:
What’s your forecast for the future?

Chip:
If we went back seven years, no one would have predicted where AI is today. The pace is only accelerating. AI will be more integrated, more pervasive, and more capable.

The shift insurers need to make is moving from pure automation (“do it faster”) to intelligent automation (“do it smarter”). To get there, they need the right foundation—clean data, modern processes, and infrastructure designed to feed AI systems.

Companies building that foundation now will be the ones able to take advantage of the next wave of AI tools and capabilities.

Andrew:
Chip, thanks for chatting with us. Enjoy the rest of ITC!

Chip:
Thank you—great conversation.

Andrew:
And thanks to everyone tuning in. We’ll see you next year here at the ITC Live Stage sponsored by Monitor.